Targeted Disruption of Ing2 Results in Defective Spermatogenesis and Development of Soft-Tissue Sarcomas

ING2 (inhibitor of growth family, member 2) is a member of the plant homeodomain (PHD)-containing ING family of putative tumor suppressors. As part of mSin3A-HDAC corepressor complexes, ING2 binds to tri-methylated lysine 4 of histone H3 (H3K4me3) to regulate chromatin modification and gene expression. ING2 also functionally interacts with the tumor suppressor protein p53 to regulate cellular senescence, apoptosis and DNA damage response in vitro, and is thus expected to modulate carcinogenesis and aging. Here we investigate the developmental and physiological functions of Ing2 through targeted germline disruption. Consistent with its abundant expression in mouse and human testes, male mice deficient for Ing2 showed abnormal spermatogenesis and were infertile. Numbers of mature sperm and sperm motility were significantly reduced in Ing2 −/− mice (∼2% of wild type, P<0.0001 and ∼10% of wild type, P<0.0001, respectively). Their testes showed degeneration of seminiferous tubules, meiotic arrest before pachytene stage with incomplete meiotic recombination, induction of p53, and enhanced apoptosis. This phenotype was only partially abrogated by concomitant loss of p53 in the germline. The arrested spermatocytes in Ing2 −/− testes were characterized by lack of specific HDAC1 accumulation and deregulated chromatin acetylation. The role of Ing2 in germ cell maturation may extend to human ING2 as well. Using publicly available gene expression datasets, low expression of ING2 was found in teratozoospermic sperm (>3-fold reduction) and in testes from patients with defective spermatogenesis (>7-fold reduction in Sertoli-cell only Syndrome). This study establishes ING2 as a novel regulator of spermatogenesis functioning through both p53- and chromatin-mediated mechanisms, suggests that an HDAC1/ING2/H3K4me3-regulated, stage-specific coordination of chromatin modifications is essential to normal spermatogenesis, and provides an animal model to study idiopathic and iatrogenic infertility in men. In addition, a bona fide tumor suppressive role of Ing2 is demonstrated by increased incidence of soft-tissue sarcomas in Ing2− /− mice.


Introduction
ING2 (inhibitor of growth family, member 2) plays pivotal roles in the regulation of cellular senescence, apoptosis, DNA damage repair, gene transcription and chromatin modification [1,2]. Our previous in vitro studies on cellular senescence suggested that ING2 functionally interplays with the p53 tumor suppressor protein in two different manners: endogenous ING2 inhibits senescence and the transcriptional repression of ING2 by p53 abrogates this inhibition [3]; and overexpressed ING2 enhances p53 acetylation and stability to induce senescence [4,5]. ING2, as a subunit of the mSin3A-HDAC1 (histone deacetylase 1) complex, specifically binds to tri-methylated lysine 4 of histone H3 (H3K4me3) via its plant homeodomain (PHD) finger and regulates gene expression through chromatin modifications in response to DNA damage [6,7]. Although these findings imply that ING2 may contribute in vivo to p53-regulated processes, as well as developmental and homeostatic processes involving chromatin regulation, the in vivo physiological roles of ING2 have not been experimentally examined.
Spermatogenesis, a series of spermatogenic cell differentiation steps from spermatogonia to mature spermatozoa in the testes, is a process tightly regulated by chromatin modifications [8,9]. Enzymes that modify histone methylation, including Meisetz (an H3K4 tri-methyltransferase) [10], G9a [a mono-and dimethyltransferase on lysine 9 of histone H3 (H3K9)] [11], and Suv39h1 and Suv39h2 (H3K9 tri-methyltransferases) [12], are essential for normal germ cell development in mice. The stagespecific acetylation profiles of several lysine residues on core histones, i.e., acetylation in spermatogonia and deacetylation during differentiation from leptotene to pachytene stages, are also critical to normal spermatogenesis in mice [8,13]. Consistently, HDAC inhibitors impaired male fertility in mice through loss of pachytene spermatocytes and increased apoptosis [14,15,16]. Such dynamic regulation of chromatin modifications during spermatogenesis is not limited to mice: the H3K4 methylation profiles during spermatogenesis were very similar between mice and nonhuman primates [17]. Misregulated histone acetylation is associated with defective spermatogenesis in humans [18], suggesting that chromatin-mediated regulation is a conserved mechanism from rodents to humans. However, no specific gene defect has been identified in humans as responsible for spermatogenic defect due to aberrant chromatin regulation [19].
In this study, our generation and characterization of Ing2deficient mice reveal that ING2 plays an essential in vivo role in mammalian spermatogenesis, which is attributed to its functional interaction with p53 and chromatin regulation. The relevance in humans is underscored by bioinformatics analysis revealing low ING2 expression in men with infertility and defective spermatogenesis. In addition, loss of Ing2 resulted in high incidence of softtissue sarcomas, particularly histiocytic sarcomas, demonstrating, for the first time, a tumor suppressor role for Ing2.

ING2 is expressed abundantly in mouse and human testes
Quantitative RT-PCR (qRT-PCR) analysis of Ing2 mRNA expression in normal mouse tissues demonstrated a tissue-specific expression pattern, with testis showing the highest level of Ing2 (Fig. 1A), as has been reported in humans [20]. Immunohistochemical (IHC) staining of human testis sections showed that the cells in seminiferous tubules expressed higher levels of ING2 protein than the interstitial cells (Fig. 1B, Fig. S1). These data indicate that germ cells are the major source of ING2 expression in mouse and human testes.

Ing2-deficient males are infertile
The postnatal growth of Ing2 2/2 mice was indistinguishable from that of their Ing2 +/+ littermates. Ing2 2/2 mice had significantly smaller testes than those of Ing2 +/+ mice throughout their life (Fig. 1C,D and Table 1, P,0.001), but did not show any gross abnormalities in other organs, including seminal vesicles, epididymides and vasa deferens (Fig. 1C, Table 1). The small-sized testes of Ing2 2/2 mice were not due to serum testosterone levels (Table 1).

Ing2 2/2 testes show degeneration of seminiferous tubules
Histological analysis showed normal germ cell development in seminiferous tubules in Ing2 +/+ testes ( Fig. 2A,B). In contrast, Ing2 2/2 testes exhibited seminiferous tubule degeneration, germ cell under-population (Fig. 2C,D) with Leydig cell hyperplasia, apoptotic cells and multinucleated giant cells (Fig. 2E). In Ing2 2/2 seminiferous tubules, spermatogonia are present in low numbers and in some tubules various developmental stages can be seen up to round spermatid. The major cell stage observed in these tubules is spermatocyte with large dense nuclei (Fig. 2D, 2B, black arrows). However, postmeiotic cell types such as round and elongated spermatids (Fig. 2B, white allow) were scarcely observed in Ing2 2/2 tubules (Fig. 2D), suggesting a spermatogenesis arrest at meiotic phase in most tubules. Observation of the epididymal wall also suggests that the cellular organization of this structure is also compromised in Ing2 2/2 mice. While mature spermatozoa were present in the epididymides of Ing2 +/+ mice, hypospermia was observed in Ing2 2/2 (Fig. 2F).
Ing2 +/+ and Ing2 2/2 testes were histologically examined throughout the mouse lifespan (Fig. S3). While Ing2 +/+ seminiferous tubules developed normally and became completely populated with all stages of germ cells by 6 to 8 weeks, most Ing2 2/2 tubules remained devoid of germ cells and/or disorganized at the same age. Analysis at 6 through 24 months of age reveals that Ing2 +/+ tubules maintained normal germ cell development. Ing2 2/2 tubules, however, underwent progressive degeneration and germ cell depletion, eventually showing a Sertoli-cell-only pathology at 24 months. When the epididymis was examined at 24 months, mature spermatozoa were present in Ing2 +/+ , while the Ing2 2/2 epididymis displayed an aspermic phenotype.
Meiotic arrest and impaired meiotic recombination in Ing2 2/2 testes DNA contents were analyzed in cells from 2-and 6-month-old Ing2 +/+ and Ing2 2/2 testes (Fig. 2G). Ing2 2/2 testes showed reduced 1N fractions (representing round and elongated spermatids) at 2 months of age, which were almost absent at 6 months. A relative increase in 2N fractions (spermatogonia and somatic cells) in Ing2 2/2 testes can be attributed to a failure to complete meiosis II as well as to the presence of somatic cells associated with Leydig cell hyperplasia and Sertoli-cell-only tubules in degenerated seminiferous tubules. An apparent loss of 4N fractions (spermatocytes) in Ing2 2/2 testes became evident at 6 months that can be attributed to premature loss of germ cells with ageing. Immunohistochemical (IHC) staining of phosphorylated histone H2AX (c-H2AX) marked DNA double-strand breaks at leptotene stage associated with meiotic recombination [21,22] in both Ing2 +/+ and , as normally expected with the completion of meiotic recombination [21,22]. In contrast, Ing2 2/2 seminiferous tubules were defective in progression to the c-H2AX-negative pachytene stage (Fig. 2J,K) and instead contained spermatocytes with abnormally accumulated c-H2AX (Fig. 2K, white arrowheads). These results are consistent with the histopathological observations described above, and indicate a failure to complete meiosis, as well as a progressive loss of germ cells with aging in Ing2 2/2 mice.
Gene expression profiles in Ing2 2/2 testes are consistent with spermatogenesis arrest Gene expression profiling by mRNA microarray analysis using whole testis RNA identified 619 genes differentially expressed (381 downregulated and 238 upregulated, P,0.001, FDR,0.04) in Ing2 2/2 testes compared with Ing2 +/+ testes (GSE18610 at http:// www.ncbi.nlm.nih.gov/geo/). Ing2 was the most downregulated gene (Table S1), confirming the validity of the experimental system. A genelist consisting of genes differentially expressed in Ing2 2/2 testes was used to query a database of gene expression profiles during mouse spermatogenesis, including spermatogonia, spermatocytes and spermatids, as well as Sertoli cells, whole seminiferous tubules, and whole testes [GermOnline, http://www. germonline.org [23]]. There was a significant correlation (nonparametric Spearman correlation P,0.0001) between gene expression in Ing2 2/2 testes and gene expression during mouse spermatogenesis. Specifically, genes differentially expressed in Ing2 2/2 testes were positively correlated with genes expressed in Sertoli cells (Spearman r = 0.72 and spermatogonia (Spearman r = 0.70), but negatively correlated with genes expressed in spermatocytes (Spearman r = 20.25) and spermatids (Spearman r = 20.74). These expression profiles further validate the conclusions from our histopathological and IHC findings that indicate failed differentiation of spermatocytes as the major defect caused by Ing2 deficiency. Student's t test, n = 5 per group. Error bars are s.e.m. (E) Semen in Ing2 +/+ and Ing2 2/2 mice. Scale bars, 100 mm. (F) Numbers of normal spermatozoa in semen during postnatal development of Ing2 +/+ and Ing2 2/2 mice. Data are mean from n = 3 (8 weeks) or n = 2 (16 and 24 weeks). Error bars are s.e.m. (G) Normal and abnormal spermatozoa from Ing2 +/+ and Ing2 2/2 mice, respectively. The leftmost panel (Ing2 +/+ ) shows mature normal spermatozoon with a characteristic, hook-shaped nucleus, straight mitochondrial sheath and straight tail. The right three panels (Ing2 2/2 ) show morphologically abnormal, immature spermatozoa, including ones with a round head and a short tail (black arrow), a large head and multiple tails (white arrow), and a tail coiled (white arrowhead). Scale bars are 50 mm in the left three panels and 25 mm in the rightmost panel. doi:10.1371/journal.pone.0015541.g001  Table S1. A group of genes involved in normal differentiation and function of spermatids or sperms, including Prss21 [24], Sly [25], Spef2 [26], Ssty2 [27] and the Speer family of testis-specific genes [28], were downregulated in Ing2 2/2 testes, consistent with impaired progression to these later stages of male germ cell development. Two downregulated genes, Asb4 (ankyrin repeat and SOCS boxcontaining protein 4) [29] and Gzmn (granzyme N) [30], are believed to be specifically involved in spermatocyte differentiation at pachytene stage, when Ing2 deficiency manifested its effect. The marked downregulation of these two genes in Ing2 2/2 testes was validated by qRT-PCR (Fig. S4). It should also be noted that several genes encoding chromatin modifying or associated proteins, including Setdb2 [31], Zfp57 [32], Suv39h2 [12], Satb2 [33], Ing3 [34] and Phf21a [35], were downregulated in Ing2 2/2 testes. In addition to the deregulation of chromatin modifications by loss of Ing2 itself, decreased expression of these proteins may possibly affect chromatin status and contribute to defective spermatogenesis in Ing2 2/2 testes. A set of somatic cellderived transcripts upregulated in testes of HDAC inhibitortreated mice [14] were not significantly changed in Ing2 2/2 testes (Table S2).
p53-dependent and independent apoptosis in Ing2 2/2 testes Because of the functional association of ING2 with p53 and the well-characterized role of p53 in regulating apoptosis in the testis [36,37,38], we next explored the effect of Ing2 deficiency on p53 expression. To this end, 8-week-old Ing2 +/+ and Ing2 2/2 testes, as well as p53 2/2 testes as a negative control, were examined for p53 protein expression. A ,2.5-fold increase in p53 protein was observed in whole testis lysates of Ing2 2/2 mice compared to agematched Ing2 +/+ mice (Fig. 4A,B). Immunohistochemical analysis of p53 protein revealed intratubular staining only in germ cells of Ing2 2/2 testes, but not Ing2 +/+ or p53 2/2 testes (Fig. 4C). No significant increase in p53 mRNA level was found in Ing2 2/2 testes (Fig.4D) suggestive of a post-transcriptional mechanism for p53 induction. PUMA, a p53-induced apoptosis effector, was also upregulated in luminal regions of the tubules in Ing2 2/2 testes (Fig. S6). These results indicate that endogenous ING2 promotes germ cell survival and differentiation. Loss of ING2 leads to p53 activation in vivo, perhaps as an indirect result of testicular degeneration or through a novel regulatory interaction between p53 and ING2.
To better characterize the tubular degeneration brought about by Ing2 deficiency, we determined the presence of apoptotic or senescent germ cells in testes at 8 weeks of age. We found that Ing2 2/2 testes had significantly increased numbers of TUNELpositive tubules and TUNEL-positive cells per tubule, compared with those in age-matched Ing2 +/+ testes (Fig. 5A, left two bars in both panels). While apoptotic germ cells were observed rarely and close to the basement membrane in Ing2 +/+ testes (Fig. 5B, Ing2 +/+ ), they occurred frequently in spermatocytes in luminal regions of the tubules (Fig. 5B, Ing2 2/2 ). Neither Ing2 +/+ nor Ing2 2/2 testes at 8 week of age showed positive staining for senescence-associated b-galactosidase, a marker of cellular senescence [39], in their seminiferous tubules (data not shown).

Reduced ING2 expression is associated with impaired spermatogenesis and male infertility in humans
To examine the expression of ING2 in men with low sperm count, morphologically abnormal sperm and/or impaired spermatogenesis, public microarray datasets available at GEO (http:// www.ncbi.nlm.nih.gov/geo) and ArrayExpress (http://www.ebi. ac.uk/microarray-as/ae) were queried using the Nextbio search engine (http://www.nextbio.com) ( Table 2). Two datasets (GSE6967 and GSE6872) comparing spermatozoa purified from semen samples of infertile teratozoospermic men versus normal fertile men [42] showed significantly lower levels of ING2 expression in the former cases. In two independent studies comparing testicular biopsies with impaired spermatogenesis classified according to Johnsen score [43] versus full spermatogenesis (GSE4797 and five comparisons in E-TABM-234) [44,45], decreased expression of ING2 was consistently associated with the spermatogenic defect. Furthermore, ING2 expression was lower in testicular biopsy specimens from non-obstructive azoospermia (NOA) patients than those from obstructive azoospermia (OA) patients (GSE9210) [46], suggesting the functional involvement of reduced expression of ING2 in human male infertility due to defective germ cell development, but not that due to a physical obstruction. Considering that germ cells are the major source of ING2 expression in testis, the finding of low ING2 expression in men with Sertoli-cell only syndrome likely reflects the fact that the spermatogenic cell types are absent from the seminiferous tubules. This association does not per se indicate a causative role for ING2 deficiency in human male infertility.  Western blot analysis of p53 protein expression using whole testis lysates from Ing2 +/+ and Ing2 2/2 testes, as well as p53 2/2 testes (negative control), at 8 weeks of age. b-actin was a loading control. Relative p53 expression levels (normalized to b-actin) based on quantitative image analysis are shown in (A) and in (B). Student's t test. n = 3 per group. Error bars are s.e.m. (C) p53 staining in testes from 8-week-old p53 2/2 , Ing2 +/+ and Ing2 2/2 mice. Black arrows indicate intratubular positive staining in Ing2 2/2 testes. Both Ing2 +/+ and Ing2 2/2 testes, as well as p53 2/2 testes to a much lesser extent, show interstitial staining (white arrowheads). Scale bars, 100 mm. (D) Real-time qRT-PCR analysis of p53 mRNA expression. Two independent primers/probe sets (assays 1 and 2) were used. The expression levels in Ing2 2/2 testes are shown as the relative values to those in Ing2 +/+ testes. Data are mean 6 s.e.m. from n = 8 (Ing2 +/+ ) or n = 14 (Ing2 2/2 ). No statistically significant difference was observed (Student's t test). doi:10.1371/journal.pone.0015541.g004

Ing2 is a tumor suppressor gene
To examine whether Ing2 deficiency affects tumorigenesis and the aging processes, we observed Ing2 +/+ (n = 22) and Ing2 2/2 (n = 28) mice for two years. There was no statistically significant difference in survival between the two genotypes ( Fig. S7, P = 0.43, Log-rank test). However, detailed histopathological analysis of both groups at time of death showed a significant change in malignant tumor spectrum. The incidence of soft tissue sarcomas was increased in Ing2 2/2 mice (P = 0.017) ( Table 3). The major tumor type observed in Ing2 2/2 mice was histiocytic sarcoma (Fig.  S8, Table 3), which showed increased incidence preferentially in males for currently unknown reasons (Table S3). Although we confirmed significantly increased frequencies of degenerated seminiferous tubules and oligozoospermic epididymis in Ing2 2/2 mice (Tables S4 and S5), pathological examinations of other organs and tissues showed that Ing2 2/2 mice had no remarkable difference from Ing2 +/+ mice in the spectrum of non-malignant lesions, except for increases in benign harderian gland adenomas and in atypical lymphoid hyperplasia of the spleen and a decrease in acinar dilation in prostate (Table S4). Ing2 2/2 mice also showed no signs of premature aging, such as hair graying, alopecia, skin atrophy, lordokyphosis, osteoporosis or cataracts. Thus, we currently conclude that Ing2 deficiency by itself does not affect aging phenotypes but enhances spontaneous formation of softtissue sarcomas.

Discussion
This study provides evidence that ING2 is an essential regulator of mammalian spermatogenesis by showing that: 1) testes express high levels of ING2 (Fig. 1A,B, Fig. S1); 2) genetic knockout of Ing2 causes a spermatogenesis defect and male infertility in mice (Figs 1C-G, 2A-F and Table 1); and 3) decreased ING2 expression is highly associated with defective spermatogenesis and male infertility in humans ( Table 2). The data indicate that the spermatogenic function of ING2 depends on both its regulatory effect on chromatin (Fig. 3) as well as its functional interaction with p53 (Fig. 4,5).
The spermatocytes in defective Ing2 2/2 seminiferous tubules underwent meiotic arrest (Fig. 2G), which was also observed in mouse models deficient for histone methyltransferases [10,11,12]. However, unlike these mouse models in which fertility was impaired in both male and female, Ing2 deficiency only affected male fertility, suggesting a distinct role for ING2 in regulating chromatin modification during mammalian germ cell development. Histopathologically, the blocked differentiation of Ing2 2/2 spermatocytes into pachytene stage (Fig. 2H-K,5A) most resembled the meiotic defect in testes in mice treated with a HDAC1 inhibitor, trichostatin-A [15]. The ability of ING2 to recruit HDAC1 to H3K4me3 [6], the dynamic regulation of H3K4me3 during zygotene-to-pachytene progression [17], and the accumulation of HDAC1 and the deacetylation of core histones in pachytene spermatocytes in Ing2 +/+ but not Ing2 2/2 mouse testes ( Fig. 3A) all suggest that the spermatogenesis defect by Ing2 deficiency is due to a disturbance of the stage-specific histone modifications coordinated by the H3K4me3-ING2-HDAC1 interaction. Our data establish ING2 as a chromatin-associated, non-enzyme protein that is critical to temporal and spatial profiles of chromatin modifications during spermatogenesis. A recent study provides further support to this hypothesis. Suberoylanilide hydroxamic acid (SAHA), an HDAC inhibitor used clinically for the treatment of cancer causes the dissociation of ING2 from the Sin3 deacetylase complex leading to de-repression of downstream genes and growth inhibition [47]. Our findings reveal a plausible mechanism by which HDAC inhibitors may disrupt deacetylase function during spermatogenesis, through disruption of Ing2bound co-repressor complexes. Consistent with this is also the finding that mice deficient on the Sirt1 protein deacetylase show spermatogenesis defects similar to Ing2-deficient mice [48]. SIRT1 is recruited by ING proteins to negatively regulate mSIN3A/ HDAC1 transcriptional repression activity [49]. It is unknown whether an altered expression of one or a few specific genes or a genome-wide change in chromatin status mediates the defects in Ing2-deficient mice, as well as in other infertile mice deficient for chromatin modifying factors. The two pachytene-specific genes identified in this study, Asb4 and Gzmn (Fig. S4), deserve further investigation. The somatic cell-derived transcripts differentially expressed in testes of trichostatin-A-treated mice [14] were not significantly affected by Ing2 deficiency (Table S2), in agreement with the findings that germ cells are the major cell type expressing ING2 (Fig. 1B, Fig. S1). Ing2-deficient mice should thus be a suitable model to study germ cell-autonomous effects of chromatin modifications on spermatogenesis. Ing2 2/2 mice exhibited phenotypic differences from mice deficient for Ing1, the founder member of the ING family, which could be explained by the difference in organ-specific expression profiles [20], the different interacting proteins [1], the different modes of functional interaction with p53 [50], and/or the different p53-independent functions [51]. ING1 expression was low in testes [20] and Ing1-deficient mice were fertile [51,52]. Whereas Ing1 2/2 mice had reduced body size [52], Ing2 2/2 mice grew normally. Furthermore, while Ing1 2/2 mice had elevated incidence of B-cell lymphomas [51,52], Ing2 2/2 mice had elevated incidence of histiocytic sarcoma (Table 3,S3). p19 Arf -deficient mice exhibit similar phenotypes to Ing2 2/2 mice, such as testicular atrophy, increased germ cell apoptosis, loss of sperm and high incidence of histiocytic sarcoma [53,54]. This most intriguing finding brings up the possibility that ING2 and p19 Arf , two p53 regulatory proteins, are involved in a common pathway that is aimed to restrict selfrenewal and ensure normal differentiation during spermatogenesis as well as prevent tumor development.
p53 regulates apoptosis in testes, which eliminates germ cells with DNA damage induced by irradiation [36,37]or spontaneously occurring during normal spermatogenesis [38]. p53-independent apoptosis also occurs during spermatogenesis in response to DNA damage and meiotic arrest [37,55]. Our comparative analysis of apoptosis in Ing2 +/+ , Ing2 2/2 , p53 2/2 and p53 2/2 Ing2 2/2 testes (Fig. 5A,B) indicates that both p53-dependent and independent mechanisms of apoptosis were activated by Ing2 deficiency, probably contributing to the elimination of developmentally arrested spermatocytes with unprocessed DNA lesions (Fig. 2J,K). The induction of p53 by Ing2 deficiency (Fig. 4) suggests that a physiological in vivo function of endogenous ING2 may be to prevent an illegitimate activation of p53 under non-stressed conditions. However, this functional link between ING2 and p53 is not a sole mechanism by which ING2 ensures normal spermatogenesis and male fertility, as indicated by the mitigated but still significant pathological changes in testes (Fig. 5C,D), the accumulation of abnormal spermatozoa (Fig. 5C) and infertility in p53 2/2 Ing2 2/2 males.
Approximately 10% of couples suffer from infertility with approximately equal contributions from men and women [56]. Meiotic arrest during spermatogenesis, as observed in Ing2 2/2 mice, is frequently associated with male infertility in humans [57]. However, because of the complexity of the spermatogenic process possibly involving thousands of different genes, a majority of male infertility cases in humans remain not understood [19]. Given a disturbed pattern of chromatin modification in defective spermatogenesis in humans [18] and decreased ING2 expression commonly associated with sperm abnormalities and pathological changes in spermatogenesis in humans (Table 2), we propose that the ING2-mediated chromatin regulation is critical to normal meiotic progression during spermatogenesis in humans as well and may be widely, if not universally, impaired by a number of different spermatogenic defects leading to idiopathic infertility in men. Genetic and epigenetic changes, as well as functional variants, of the ING2 gene itself also deserve investigation in infertile men. This study also has significant implications in infertility in men with other diseases, in particular malignant tumors. Patients with testicular cancers are frequently infertile [58,59], with the degeneration of peritumoral, non-cancerous seminiferous tubules [18,58]. HDAC inhibitors are under clinical trial in patients with various types of solid and hematologic malignancies [60,61]. Spermatogenesis is a process highly sensitive to DNA damage by common chemotherapy and radiotherapy [58,62] and male infertility is a major quality-of-life issue in cancer survivors [58,63]. We expect that the Ing2 2/2 mice generated and characterized in this study can be a model system to study idiopathic and iatrogenic male infertility in humans.

Materials and Methods
Generation of Ing2 2/2 mice This mice study was approved by the ACUC, NCI-Frederick and all the guidelines were followed for the study. To construct the targeting vector for Ing2, a BAC clone containing the Ing2 locus was isolated from a 129-mouse genomic library (Genome Systems Inc, St.Louis, MO). A 6.2-kilobase (kb) HindIII-HindIII fragment encompassing from the first intron to the second intron was cloned into the HindIII site of the pBS-SK(-)Pst(-) vector, pBluescript SK(-) (Stratagene, La Jolla, CA) with PstI site removed. A 1.7-kb HindIII-HindIII fragment within the second intron was cloned into the HindIII site of the pLoxpneo vector (a gift from Dr. Chuxia Deng) [64]. A single loxp site was inserted into the PstI site at the first intron within the 6.2-kb fragment cloned in the pBS-SK(-) Pst(-). The resulting plasmid was cleaved with NotI and XhoI and inserted into the NotI and XhoI sites of the pLoxpneo containing the 1.7-kb fragment. The final targeting construct was designated pLoxpneoIng2 (Fig. S1a). The targeting construct was linearized and electroporated into ES cells derived from 129/Sv mice. G418resistant colonies were selected and expanded. The ES clones with correct targeting events were identified by Southern blot and PCR. Heterozygous mice with the targeted allele (Ing2 flox2neo/+ ) were crossed with each other, resulting in homozygous mice (Ing2 flox2neo/flox2neo ). These homozygous mice were crossed with EIIa-Cre [65] mice to remove DNA fragments between two loxp sites. This led to the generation of three types of offspring with a deletion of the neomycin resistance cassette (Ing2 +/flox ), the exon 2 (Ing2 +/neo ), or both (Ing2 +/2 ). The identified Ing2 +/2 mice were then bred with C57BL/6J mice to segregate the EIIa-Cre transgene. Ing2 +/2 mice without EIIa-Cre were mated with each other, leading to the generation of Ing2 knockout mice (Ing2 2/2 ).

Quantitative real-time RT-PCR
Total RNAs were extracted from the whole testes of 8-week-old mice using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Five micrograms of total RNA were used for the synthesis of first strand cDNA using the SuperScript III First Strand cDNA Synthesis Kit (Invitrogen). To examine ING2 expression in various mouse organs, BD MTC Mouse Panel III was purchased from BD Biosciences (San Jose, CA). Realtime RT-PCR analysis was performed using ABI prism 7900 (Applied Biosystems, Foster City, CA) with Taqman Gene Expression Assays, purchased from Applied Biosystems; Ing2 (Mm00469833_m1), p53 (Mm00441964_g1, assay 1 in this study), p53 (Mm01337166-mH, assay 2 in this study), Asb4 (Mm00480830_m1), and Gzmn (Mm00461850_m1). GAPDH (Applied Biosystems, Mm99999915_g1) was used as internal control. Normalized or relative gene expression was calculated using the equation 2 2DCt or 2 2DDCt , respectively, where DCt = Ct(Gene)2Ct(GAPDH).

Gross anatomical and histological examinations
Gross examination of organ weight and morphology on all mice necropsied at 2 weeks to 2 years of age. For routine histopathological examination all tissues were formalin-fixed, embedded in paraffin, sectioned at 5 mm and stained with hematoxylin and eosin (H&E).

TUNEL assay
For terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) analysis, formalin-fixed sections were deparaffinized, rehydrated, and pretreated with proteinase K. Apoptotic cells were detected using DeadEnd Colorimetric TUNEL system kit (Promega, Madison, WI) according to the manufacturer's instructions.

Sperm counts and motility
For each mouse (8-, 16-and 24-week-old) whole epididymis and vas deferens were harvested, cut into 2-mm-long pieces, resuspended in 1 ml of buffer containing 75 mM NaCl, 24 mM EDTA and 0.4% bovine serum albumin. Pieces and sperm fluid were homogenized at 32uC for 10 min to dissociate somatic cells. Sperm cells remaining as a monodispersed suspension were counted on a hemocytometer.

Testosterone assay
Blood was taken from 8-and 26-week-old male mice (five in each group) that were being housed singly, in separate cages. Female mice were being housed in separate cages of the same room at the time of blood sampling. Serum testosterone level was measured using ELISA kit (Alpha Diagnostic International, San Antonio, TX) according to the manufacturer's instructions.

Flow cytometry
Testes were excised from 2-and 6-month-old mice (three mice in each group), decapsulated and crushed through 20-gauge needles and 70-mm cell strainers (BD Biosciences) in phosphatebuffered saline. Cells (2610 6 ) were treated with RNase and stained with propidium iodide using a Cycle Test Plus DNA reagent kit (Becton Dickinson, Franklin Lakes, NJ). All fluorescence-activated cell sorting data were analyzed using CELL Quest (version 3.3; Becton Dickinson).

mRNA microarray analysis
Total RNA samples were isolated from whole testes of Ing2 +/+ (n = 3) and Ing2 2/2 (n = 5) mice at 2-3 months of age using Trizol (Invitrogen). Microarray analysis was performed using Affymetrix platform (Mouse Gene 1.0 ST Array). Data was RMA-normalized using Affymetrix Expression Console, annotated according to NetAffx release: 28 (2009-03-16) and summarized at the gene level. Resulting normalized and annotated probesets were imported into BRB Array Tools (http://linus.nci.nih.gov/BRB-ArrayTools.html). A Class Comparison analysis, based on a Twosample T-test using the random variance model, yielded 1046 probesets differentially expressed between Ing2 +/+ and Ing2 2/2 testes (P,0.001, FDR,0.04). Of those, 300 were not associated with a named gene and were excluded from further analysis. After further filtering for multiple occurrences of a gene ID, 619 unique genes remained, of which 381 were downregulated and 238 were upregulated.

Bioinformatics analyses
To determine a cell type-specific expression of ING2 in mouse testis, ING2 expression values were extracted from the public Gene Expression Omnibus data repository (GEO, http://www. ncbi.nlm.nih.gov/geo/). A microarray dataset of testes from C57BL/6 mice (accession number GDS409) was interrogated. The expression values for ING2 probesets were identified within the normalized data files available at GEO, and they were subsequently plotted to generate graphical displays.
To examine whether ING2 expression levels are associated with spermatogenic defects and male infertility in humans, we used the datasets available at GEO and ArrayExpress (http://www.ebi.ac. uk/microarray-as/ae), which encompass microarray analysis of testes from men with low sperm count, morphologically abnormal sperm and/or impaired spermatogenesis. Meta-analysis was performed using the Nextbio database of curated and processed high-throughput data (http://www.nextbio.com) by querying for the search term ''ING2'' and filtering for ''male infertility'' or ''azoospermia''. The analysis output consisted of individual microarray study results that showed differential ING2 expression for the tested comparisons in each dataset. For each microarray study, the fold change and p-value of ING2 for the comparison were indicated.
To compare between the existing datasets of spermatogenesisassociated profiles of gene expression and the differentially expressed genes identified by the mRNA microarray analysis in this study, the gene symbols were compiled into two genesets: upregulated (238) or downregulated (381) in Ing2 2/2 testes. These genesets were separately uploaded to GermOnline (http://www. germonline.org [23]), a database of microarray expression profiling data from experiments relevant for the mitotic and meiotic cell cycle, gametogenesis and fertility, from which corresponding expression values were extracted. Expression intensities of 98 and 138 genes were extracted from the upregulated and downregulated genesets, respectively. Expression intensities in Sertoli cells, spermatogonia, spermatocytes, spermatids and tubules were normalized to total testis and plotted separately for each geneset.

Analysis of spontaneous tumors
Life span and spontaneous tumor incidence were determined in Ing2 +/+ and Ing2 2/2 mice. Mice with visible tumors and moribund mice showing weight loss or difficulties in moving were sacrificed upon detection. The remaining surviving mice were sacrificed at 24 months of age. Gross anatomical and histological examinations were performed as described above.

Human subjects
The histology of Mouse testis was compared with human testis obtained from the Armed Forces Institute of Pathology at Water Reed Army Medical Center (AFIP). Use of the human organs was approved by the NIH Office of Human Subjects Research (OHSR) which issued an IRB Exemption (#4545) for the study. Immunohistochemical (IHC) staining of ING2 protein in a normal human testis section was done at the AFIP at Water Reed Army Medical Center.  Figure S3 Histological analysis of testes from Ing2 +/+ and Ing2 2/2 mice at different ages. Testes (T) at 2 weeks to 24 months of age, as well as epididymis (E) at 24 months of age, were examined by H&E staining. Scale bars, 200 mm. (TIF) Figure S4 Real-time qRT-PCR analysis of Asb4 and Gzmn mRNA expressions. The expression levels in Ing2 2/2 testes are shown as the relative values to those in Ing2 +/+ testes.