• Open Access

O-GlcNAcase is essential for embryonic development and maintenance of genomic stability

Authors

  • Yong Ryoul Yang,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    2. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Minseok Song,

    1. Department of Psychiatry, Weill Cornell Medical College, New York, NY 10065, USA
    Search for more papers by this author
  • Ho Lee,

    1. Cancer Experimental Resources Branch, National Cancer Center, Goyang-si, Gyeonggi-do 410-769, Republic of Korea
    Search for more papers by this author
  • Yoon Jeon,

    1. Cancer Experimental Resources Branch, National Cancer Center, Goyang-si, Gyeonggi-do 410-769, Republic of Korea
    Search for more papers by this author
  • Eun-Jeong Choi,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Hyun-Jun Jang,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    2. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Hyo Youl Moon,

    1. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    2. School of Interdisciplinary Bioscience and Bioengineering, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea
    Search for more papers by this author
  • Ha-Young Byun,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    2. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Eung-Kyun Kim,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    2. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Dae Hyun Kim,

    1. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Mi Nam Lee,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Ara Koh,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Jaewang Ghim,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Jang Hyun Choi,

    1. Department of Cancer Biology and Division of Metabolism and Chronic Disease, Dana-Farber Cancer Institute, Boston, MA 02115, USA
    Search for more papers by this author
  • Whaseon Lee-Kwon,

    1. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author
  • Kyong Tai Kim,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Sung Ho Ryu,

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    Search for more papers by this author
  • Pann-Ghill Suh

    1. Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea
    2. School of Nano-Biotechnology and Chemical Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea
    Search for more papers by this author

Pann-Ghill Suh, Division of Molecular and Life Science, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Republic of Korea. Tel.: +82 52 217 2621; fax: +82 52 217 2609; e-mail:pgsuh@unist.ac.kr

Summary

Dysregulation of O-GlcNAc modification catalyzed by O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) contributes to the etiology of chronic diseases of aging, including cancer, cardiovascular disease, type 2 diabetes, and Alzheimer’s disease. Here we found that natural aging in wild-type mice was marked by a decrease in OGA and OGT protein levels and an increase in O-GlcNAcylation in various tissues. Genetic disruption of OGA resulted in constitutively elevated O-GlcNAcylation in embryos and led to neonatal lethality with developmental delay. Importantly, we observed that serum-stimulated cell cycle entry induced increased O-GlcNAcylation and decreased its level after release from G2/M arrest, indicating that O-GlcNAc cycling by OGT and OGA is required for precise cell cycle control. Constitutively, elevated O-GlcNAcylation by OGA disruption impaired cell proliferation and resulted in mitotic defects with downregulation of mitotic regulators. OGA loss led to mitotic defects including cytokinesis failure and binucleation, increased lagging chromosomes, and micronuclei formation. These findings suggest an important role for O-GlcNAc cycling by OGA in embryonic development and the regulation of the maintenance of genomic stability linked to the aging process.

Introduction

O-GlcNAcylation culminates with UDP-GlcNAc transfer to target proteins in the hexosamine biosynthetic pathway (HSP). O-GlcNAcylation is a post-translational modification that influences fundamental functions of proteins by regulating protein–protein interactions, altering protein stability, and changing protein activity (Torres & Hart, 1984; Hart et al., 2007, 2010). O-GlcNAc addition and removal from the Ser and Thr residues of proteins is catalyzed by O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA), respectively. OGA, a nucleocytoplasmic β-N-acetylglucosaminidase, is a bifunctional protein containing a glycosidase domain at the N-terminus and a putative histone acetyltransferase domain at the C-terminus (Toleman et al., 2004). OGA is mapped to the chromosomal region that is associated with Alzheimer’s disease and the development of type 2 diabetes in a Mexican population (Lehman et al., 2005). In addition, many studies have directly linked dysregulation of O-GlcNAc cycling to chronic diseases of aging including diabetes, cardiovascular disease, neurodegenerative disorders, and cancer (Hart et al., 2010). Although the cellular and physiological roles of O-GlcNAc cycling have been extensively investigated, how OGA activity is regulated and what the physiological effects are when OGA is dysregulated in vivo remain unclear.

Aging is a multifaceted process characterized by genomic instability and epigenetic changes. Maintenance of genomic stability requires accurate segregation of sister chromosomes during mitosis (Draviam et al., 2004; Holland & Cleveland, 2009). OGT is co-localized to the midbody with OGA, kinase, and phosphatase, and it regulates the post-translational status of vimentin during mitosis (Slawson et al., 2005, 2008). Correlatively, a combined glycoproteomic and phosphoproteomic approach identified O-GlcNAcylation and phosphorylation sites on various spindle and midbody proteins, suggesting that the cross talk between O-GlcNAcylation and protein phosphorylation is crucial to the regulation of cell division (Wang et al., 2010). A recent study showed that OGT glycosylates the cell cycle regulator host cell factor-1 (HCF-1) and cleaves HCF-1 to regulate mitosis (Capotosti et al., 2011). Regarding epigenetic changes, histones are O-GlcNAcylated within the nucleosome core, suggesting a crucial role for O-GlcNAc in chromatin remodeling and epigenetic regulation (Sakabe & Hart, 2010; Sakabe et al., 2010). Thus, we speculate that precise control of O-GlcNAc cycling by OGA and OGT is required for the maintenance of genomic stability and epigenetic regulation associated with aging. Moreover, we found that aged tissues have dysregulated O-GlcNAcylation and O-GlcNAc cycling enzymes.

To elucidate the cellular and physiological roles of OGA associated with dysregulation of O-GlcNAcylation, we generated OGA−/− mice. In our study, we found that OGA homozygous null mice showed perinatal lethality with developmental delay. We observed that O-GlcNAcylation level fluctuated during the cell cycle and constitutively elevated O-GlcNAcylation by OGA disruption resulted in an aberrant cell cycle. OGA−/− mouse embryonic fibroblasts (MEFs) displayed genomic abnormalities, followed by cytokinesis defects, and became senescent. Together, these findings indicate that OGA dysregulation might contribute to the mammalian aging process and affect development of age-related chronic diseases.

Results

Increased O-GlcNAcylation and decreased O-GlcNAc cycling enzymes in aged tissues

To investigate how O-GlcNAcylation level is altered in natural aging, we analyzed the levels of O-GlcNAcylation and O-GlcNAc cycling enzymes in various tissues of 4-month-old and 20-month-old mice. Notably, the relative O-GlcNAcylation level of older mice was significantly higher than that of younger mice in diverse tissues including brain, lung, skin, thymus, testis, and liver. The protein levels of O-GlcNAc cycling enzymes, OGT and OGA, decreased in tissues of older mice (Fig. 1A–D and data not shown). Although lungs from older mice showed no difference in OGA protein level as compared to lungs from younger mice, O-GlcNAcyltion was elevated (Fig. 1B). This result indicates that the consequent elevation of O-GlcNAcylation might be resulted from increased OGT activity or decreased OGA activity, suggesting the importance of the OGT/OGA balance in normal conditions. We suggest that O-GlcNAc cycling is dysregulated in aged tissue, which is caused by abnormality of the OGT/OGA balance. These results raise the possibility that increased O-GlcNAcylation is associated with the normal aging process.

Figure 1.

 Increased total O-GlcNAcylation and decreased OGA and OGT protein levels in aged tissue. (A–D) Western blot analysis of OGT, OGA, and O-GlcNAc levels in brain (A), lung (B), skin (C), and thymus (D) of old and young mice. Densitometry was performed on immunoblots. The ratio of O-GlcNAc to β-actin was determined. Error bars represent the S.D. (= 3) and **< 0.005, **< 0.05 (Student’s t-test).

Generation of OGA−/− mice and constitutively elevated O-GlcNAcylation in OGA−/− embryos

To examine whether the increased O-GlcNAcylation had cellular and physiological relevance to aging, we generated OGA-deficient mice with a gene-trapped embryonic cell line [YTC085; (Stryke et al., 2003)]. 5′-RACE analysis of the fusion transcript confirmed insertion of the gene trap vector in the first intron of OGA and identified the insertion site (Fig. S1A,B). OGA disruption was confirmed by polymerase chain reaction (PCR) using mouse tail-derived genomic DNA (Fig. S1C). OGA expression was visualized by whole-mount LacZ staining of OGA+/− embryos and placentas, confirming the insertion of the gene trap beta-galactosidase reporter allele (Fig. S1D). Embryo sagittal sections and placenta radial sections stained with X-gal at embryonic day (E) 14.5 revealed that OGA expression was particularly high in the brain and in the labyrinth structure of the placenta (Fig. S2A,B). OGA disruption resulted in constitutively elevated O-GlcNAcylation in OGA−/− embryos (Fig. 2A).

Figure 2.

OGA−/− mice die within 1 day of birth and show significantly decreased intrauterine growth. (A) Immunoblot analysis confirmed the loss of OGA protein in the OGA+/+ and OGA−/− whole embryo lysates. (B) Morphology of E12.5, E14.5, E18.5, and neonate littermates. (C) Weight of WT and OGA−/− embryos from different gestational periods and neonates. Males and females were pooled. Ten or more embryos per genotype were analyzed. Error bars represent the SD and **< 0.005, *< 0.02 (Student’s t-test). (D) Sagittal sections of OGA+/+ and OGA−/− E14.5 embryos stained with hematoxylin and eosin. (E) Hematoxylin and eosin staining of OGA+/+ and OGA−/− lungs of E18.5 embryos and P0 neonates (a, pre-alveolar spaces; br, bronchiole).

Perinatal lethality of OGA knockout mice

To obtain OGA−/− offspring, we crossed male heterozygotes with female heterozygotes. In more than 100 offspring, we failed to identify the homozygous null mutation at the time of weaning, suggesting that OGA loss causes either embryonic or neonatal lethality. We analyzed OGA−/− embryos at different gestational stages and found that OGA−/− embryos had stunted growth and that mutant neonates died within 1 day of birth (Table 1). OGA−/− embryos were significantly reduced in size and body weight compared with wild-type (WT) embryos (Fig. 2B,C). However, histological analysis of OGA−/− embryos did not exhibit gross defects, and we failed to observe any anatomical abnormalities at E14.5 (Fig. 2D). To clarify the cause of neonatal lethality, histological analysis was performed. Lungs from WT and OGA+/− mice floated in water, while lungs dissected from OGA−/− newborns were dense and sank. As expected, lung histology of OGA−/− newborns revealed that the alveolar space was markedly reduced (Fig. 2E). Together, these results show that OGA disruption leads to embryonic developmental delay, indicating an important role of OGA in development.

Table 1.   Genotype of offspring from OGA+/− × OGA+/− mating after eight generations of backcrossing with C57BL/6 wild-type animals
Developmental stageNo. of offspring with OGA genotype of:
+/++/−+/−Resorbed
E10.581270
E11.591692
E12.591364
E13.51632127
E14.51830147
E15.5101878
E16.5122184
E17.5101583
E18.51526131
Prenatal total (406) (%)1071838436
(26.10)(44.63)(20.49)(8.78)
P0489022 
P1581310 
Prenatal total (349)10622122 

Proliferative defects in OGA−/− MEFs

Reduced body size of whole OGA−/− embryos led us to examine whether the developmental delay could be caused by a cell proliferation defect. To determine whether OGA disruption caused the proliferative defect, we used MEFs isolated from mid-gestation embryos [13.5 days post-conception (dpc)]. We compared cell growth rates at passages (P) 4 and 7 in OGA+/+ and OGA−/− MEFs. OGA−/− MEFs exhibited a reduced growth rate compared with WT MEFs (Fig. 3A). OGA knockdown by shRNA also resulted in reduced cell growth in HEK293 cells (Fig. S3A,B). As cellular senescence was characterized in vitro by a decline in growth rate, OGA−/− MEFs exhibited reduced population-doubling time compared with OGA+/+ MEFs (Fig. 3B). In addition, enlarged and flattened cell morphology was observed in OGA−/− MEFs (data not shown), which are established features of senescence (Dimri et al., 1995). Previous studies have shown that increased O-GlcNAcylation leads to apoptosis (Webster et al., 2009; Rajamani & Essop, 2010). Therefore, we asked whether the reduced growth rate was caused by increased cell death. However, annexin-V staining demonstrated that the decreased cell growth of OGA−/− MEFs was not because of apoptosis (Fig. 3B).

Figure 3.

OGA−/− MEFs exhibit reduced growth rates. (A) Growth curves of OGA+/+ and OGA−/− MEFs (passages 4 and 7). Error bars represent the S.D. (= 3). (B) Growth curves were determined by a 3T3 protocol. OGA+/+ and OGA−/− MEFs were seeded on 6-cm culture dishes at a density of 105 MEFs every 3 days. (C) The population of apoptotic cells was determined by annexin-V–FITC and propidium iodide (PI) staining. A representative dot plot (left panel) and the mean (right panel) and SD from three independent experiments are presented.

Elevation of O-GlcNAcylation in serum-stimulated cell cycle entry

Reduced growth can be attributed to cell cycle defects. Thus, we wondered how O-GlcNAcylation changed during cell cycle progression. We synchronized WT MEFs in G0 phase by serum starvation and extracted proteins from cells harvested every 4 h after serum stimulation. To dissect cell cycle progression, we analyzed cyclin B1, A, and E levels. The O-GlcNAcylation level significantly increased at 20–24 h post-release, consistent with OGT expression. However, OGA expression was not altered during the cell cycle, suggesting that OGA might be regulated by its enzymatic activity. In accordance with cyclin expression patterns, OGT expression and O-GlcNAcylation were cell cycle-dependent (Fig. 4A). OGT expression level was high in late S phase and peaked at the M phase. The same patterns of OGT expression were confirmed in HEK293 cell lines (Fig. 4B). To further confirm the cell cycle-dependent O-GlcNAcylation and OGT expression, we next synchronized WT MEFs in the mitotic phase using nocodazole, an antimitotic drug. We observed similar results showing that the OGT protein level and O-GlcNAcylation peaked at mitotic phase and decreased after release from nocodazole (Fig. 4C). Strikingly, OGT levels and cyclin B1 were similarly altered. Together, these results suggest that O-GlcNAcylation and its cycling enzymes might be under precise control to ensure successful mitosis.

Figure 4.

 O-GlcNAcylation fluctuates with OGT expression during cell cycle progression. (A) WT MEFs were arrested in G0 by serum starvation and then stimulated by serum. Total lysates were analyzed at the indicated time point after serum stimulation. Immunodetection in total lysates of the indicated cyclins, O-GlcNAc cycling enzymes, and O-GlcNAc was conducted. (B) The same experiment was performed in HEK293 cells. (C) The indicated proteins from total lysates of WT MEFs at the indicated time points after release from nocodazole were analyzed by Western blotting. Actin was used as a loading control.

Mitotic defects in OGA−/−MEFs

To investigate how constitutively increased O-GlcNAcylation by OGA knockout affected the cell cycle, we next performed flow cytometry analysis with OGA+/+ and OGA−/− MEFs. Flow cytometry analysis of DNA content revealed no significant difference between the OGA−/− MEFs and WT MEFs during the early passages (P2). However, during the late passage (P4), OGA−/− MEFs showed increased 4N and decreased 2N fractions compared with OGA+/+ MEFs, indicating a partial defect in these stages of the cell cycle (Fig. 5A,B). In addition, flow cytometry analysis of BrdU/PI-stained MEFs showed reduced amounts of G1 and S phase, while G2/M-phase cells increased in the OGA−/− MEFs (Fig. 5C). Downregulation of mitotic regulators such as Aurora and polo-like kinase has been shown to lead to defective mitosis (Kunitoku et al., 2003; Glotzer, 2005; Raouf et al., 2005). Therefore, we examined the protein levels of mitotic regulators and cyclins in OGA−/− MEFs. We found that the protein levels of mitotic regulators such as Aurora Kinase B, cyclin B1, and cdc2 decreased in OGA−/− MEFs, suggesting that the OGA−/− MEFs with 4N DNA have increased nonmitotic (4N-G1) cells. As we observed that OGT was concomitantly changed with cyclin B1 during mitosis in WT cells (Fig. 4A–C), OGT was also downregulated in OGA−/− MEFs (Fig. 5D). Together, these results suggest that precise control of O-GlcNAcylation is required for cell cycle progression.

Figure 5.

OGA−/− MEFs show mitotic defects and downregulation of mitotic regulators. (A) DNA content profile of asynchronous OGA+/+ and OGA−/− MEFs (P2 and P4). (B) The percentage of cells (P4) in each phase of the cell cycle is shown. Error bars represent the S.D. (= 3) and **< 0.005, *< 0.05 vs. OGA+/+ MEFs (Student’s t-test). (C) Cell cycle status of MEFs (P4) analyzed by FACS after BrdU and PI staining. (D) Western blot analysis of the expression levels of OGT, cdc2, AKB, cyclin B1, cyclin A, cyclin D, and cyclin E. Anti-β-actin antibody was used as loading control.

Defects in the maintenance of genomic stability

We observed an increase in the percentage of multinucleated cells (>4n) in OGA−/− MEFs, suggesting gain and loss of chromosomes (Fig. 5B). These results prompted us to examine the abnormal nuclear morphologies in OGA−/− MEFs. The WT and OGA−/− MEFs were fixed and stained with Hoechst for nuclei visualization. We observed frequent enlarged nuclei and a high incidence of multinuclear cells in OGA−/− MEFs as compared to OGA+/+ (Fig. 6A). OGA−/− MEFs contained a significantly higher number of polynucleated cells (24% of OGA−/− cells vs. 9%OGA+/+ cells) and micronuclei (15% of OGA−/− cells vs. 4.8%OGA+/+ cells) compared with WT MEFs (Fig. 6B,C). Micronuclei originate from lagging chromosomes and are stably maintained in cells. We next examined whether OGA loss led to mitotic abnormalities, including lagging chromosomes. Indeed, we observed more anaphase figures with lagging chromosomes in OGA−/− MEFs than in OGA+/+ MEFs (Fig. 6D). Quantification of the lagging chromosomes is shown Fig. 6E. These results suggest that OGA loss can cause chromosome aberrations marked with nuclear enlargement and aneuploidy. To further confirm the genomic instability in OGA−/− MEFs, we performed chromosome counts on colcemid-arrested metaphase spreads from OGA+/+ and OGA−/− MEFs at P5. In OGA−/− MEFs, a higher percentage of cells exhibited abnormal chromosome numbers than in OGA+/+ MEFs. These results indicate that the loss of OGA can lead to a loss or gain of a single chromosome (Fig. 6F,G).

Figure 6.

 OGA loss causes genomic instability, and OGA is required for completion of cytokinesis. MEFs were fixed and stained with Hoechst. (A) OGA−/− MEFs had higher numbers of binuclear cells and larger mononuclei and micronuclei compared with OGA+/+ MEFs (arrows indicate micronuclei, arrowheads denote binuclear cells). The percentage of polynucleated cells (B) and micronuclei (C) in OGA+/+andOGA−/− MEFs. (D) OGA−/− MEFs with a lagging chromosome in anaphase (arrows highlight a lagging chromosome). (E) Quantification of lagging chromosomes in OGA+/+ and OGA−/− MEFs. (F) Representative photomicrographs of Giemsa-stained metaphase spreads of OGA+/+andOGA−/− MEFs (P5). (G) Metaphase spreading profiles of OGA+/+ and OGA−/− MEFs. At least 80 cells were analyzed per genotype per experiment. In all panels, error bars represent the SD and **< 0.005, *< 0.02 (Student’s t-test). (H and I) Time-lapse DIC images of OGA+/+ (H) and OGA−/− (I) MEFs. OGA−/− MEFs displayed cytokinesis failure, resulting in multinucleation (arrows indicate nuclei, time points are in minutes), with scale bars of 50 μm. (J) Frequency of OGA+/+ and OGA−/− MEFs that failed to complete cytokinesis.

Cytokinesis failure in OGA−/− MEFs

Polyploidy can result from cytokinesis failure, which subsequently leads to chromosome instability (Stenmark & Sagona, 2010). OGT and OGA are localized to the midbody and regulate mitotic regulators during cytokinesis (Slawson et al., 2008; Wang et al., 2010). Therefore, we next examined whether the OGA loss impaired cytokinesis. OGA−/− MEFs were monitored by time-lapse differential interference contrast (DIC) microscopy. As expected, the WT MEFs underwent normal cytokinesis. Forty-seven of 53 OGA+/+ MEFs exhibited normal cell cleavage and complete separation of the two daughter cells (Fig. 6H,J). In contrast, 18 of 53 OGA−/− MEFs failed to complete cytokinesis and resulted in binucleated cells (Fig. 6I,J). OGA−/− MEFs entered mitosis normally. As shown in Fig. 6I, however, OGA−/− MEFs spent a much longer time in mitotic cell rounding during mitosis compared with WT MEFs.

Cells deficient in mitotic checkpoint regulators commonly display higher rates of polyploidy (Holland & Cleveland, 2009). To assess the possibility of mitotic checkpoint defects in OGA−/− cells, we treated WT and OGA−/− MEFs with nocodazole and analyzed the mitotic defect. However, propidium iodide (PI) staining after nocodazole treatment did not reveal a difference in the fraction of 4N and 8N DNA content between WT and OGA−/− MEFs (Fig. S4A). We also obtained the same results from the knockdown of OGA with shRNA (Fig. S4B). These observations suggest that OGA is necessary for successful completion of cytokinesis.

Together, these findings support a critical role for O-GlcNAc cycling in the maintenance of genomic stability. These results also indicate that OGA dysfunction might contribute to genomic instability linked to the aging process, as we observed increased O-GlcNAcylation and decreased protein levels of OGA and OGT in natural aging.

Discussion

In this study, we found higher levels of O-GlcNAcylation and lower levels of O-GlcNAc cycling enzymes in older mice compared with younger ones in diverse tissues. We investigated how constitutively elevated O-GlcNAcylation affects physiological processes and cellular functions by generating OGA knockout mice. OGA homozygous null embryos exhibited delayed embryonic development and died soon after birth as a result of respiratory failure. OGA−/− MEFs had a reduced growth rate that led to senescence caused by mitotic defects. We also showed that OGA loss led to abnormal chromosome segregation and frequently failed to complete cytokinesis, which ultimately caused aneuploidy. These cellular defects are likely causes of the delayed OGA−/− embryo growth and consequent death.

It has studied the important roles of O-GlcNAc cycling during cell cycle. Inhibition of OGA or overexpression of OGT delays M-phase progression in mammalian cells (Slawson et al., 2005). Importantly, O-GlcNAc level is elevated in M phase during Xenopus laevis oocytes maturation (Lefebvre et al., 2004; Dehennaut et al., 2009). Consistently, dysregulation of O-GlcNAc cycling impairs cell cycle progression in Xenopus laevis oocytes (Dehennaut et al., 2007). We observed that the O-GlcNAcylation level increased at late S phase and peaked at M phase, concomitant with increased OGT levels in mammalian cells. O-GlcNAcylation decreased during mitotic exit, which is controlled by OGA enzymatic activity (Fig. 4A–C). Thus, OGA disruption may profoundly affect multiple mitotic signaling circuits. Consequently, OGA−/− MEFs failed to maintain genomic stability, followed by abnormal chromosome segregation and cytokinesis failure. Similarly, a previous study shows that increased O-GlcNAc level by OGT overexpression leads to polyploid nuclei (Slawson et al., 2005). These results suggest that increased O-GlcNAcylation by OGA disruption or OGT overexpression causes defective mitosis.

Many mitotic regulators of O-GlcNAcylation have been identified. OGA and OGT localizes in a transient complex with vimentin (a cytoskeletal protein), Aurora kinase B (AKB), and protein phosphatase 1 (PP1) at the midbody during cytokinesis, which cooperatively regulates the post-translational status of proteins (Slawson et al., 2008). Nup153 and EMSY, which are required for nuclear envelope breakdown and maintenance of genomic stability, respectively, are also O-GlcNAcylated during M phase (Favreau et al., 1996; Hughes-Davies et al., 2003; Raouf et al., 2005; Ullman et al., 2009; Wang et al., 2010). Furthermore, nuclear mitotic apparatus protein 1 (NuMA1) plays an important role in spindle pole formation and is O-GlcNAcylated at M phase (Silk et al., 2009; Yokomori et al., 2009; Wang et al., 2010). OGT overexpression causes inhibitory phosphorylation of CDK1 and consequently reduces the phosphorylation of CDK1 target proteins (Wang et al., 2010). However, how O-GlcNAcylation controls specific mitotic regulators during mitosis has not been demonstrated. These data suggest that the precise control of O-GlcNAcylation on mitotic regulators is critical for mitosis.

In addition to genomic instability, epigenetic changes are a critical component of aging in all eukaryotes. The potential roles of O-GlcNAcylation in epigenetic regulation have been investigated. O-GlcNAcylation occurs on nuclear pores and many nuclear and chromatin-associated proteins. Many transcription factors are O-GlcNAcylated, which affects their transcriptional activity (Ozcan et al., 2010). In addition, O-GlcNAc cycling can affect the activity of RNA polymerase II (Kelly et al., 1993; Hart & Comer, 2001). Strikingly, recent studies have shown that histones H2A, H2B, and H4 are O-GlcNAcylated within the nucleosome core, implicating an important role for O-GlcNAc in chromatin remodeling and epigenetic regulation (Sakabe & Hart, 2010; Sakabe et al., 2010). Correlatively, the ogt-1 and oga-1 null mutant alleles of Caenorhabditis elegans show that disruption of O-GlcNAc cycling leads to age-related transcriptional changes (Love et al., 2010). These results also support the close relationship between the aging process and dysregulation of O-GlcNAcylation. This evidence and our results suggest that increased O-GlcNAcylation by aberrant regulation of O-GlcNAc cycling enzymes might lead to epigenetic changes that contribute to natural aging.

Because genomic instability is a hallmark of most cancers, our results raise the possibility that OGA or OGT is dysfunctional in cancer. Consistent with this idea, dysregulation of O-GlcNAcylation and aberrant protein levels of O-GlcNAc cycling enzymes are observed in tumors and in cancer cell lines. In breast tissue or thyroid cancers, decreased O-GlcNAc levels and increased OGA enzymatic activity occur (Slawson et al., 2001; Krzeslak et al., 2010). In contrast, however, many studies have recently shown increases in the O-GlcNAc levels and the protein levels of O-GlcNAc cycling enzymes in cancers. Lung and colon tissue exhibited increased O-GlcNAcylation and protein levels of O-GlcNAc cycling enzymes (Gu et al., 2010; Yu et al., 2011). Moreover, in patients with chronic lymphocytic leukemia, O-GlcNAcylated proteins increased, and the protein levels of O-GlcNAc cycling enzymes also increased (Shi et al., 2010). In breast cancer cell lines, O-GlcNAc levels are elevated, and the OGT level also increases compared with less aggressive breast cancer lines. Importantly, OGT knockdown in breast cancer cells inhibits tumor growth both an in vitro and in an in vivo mouse model (Caldwell et al., 2010). OGT deficiency in a mammalian model reveals that OGT is essential for stem cell viability (Shafi et al., 2000). Moreover, the tissue-specific OGT knockout using floxed mice leads to T-cell apoptosis and fibroblast growth arrest (O’Donnell et al., 2004). We show here that OGA−/− MEF exhibited reduced growth with a decreased OGT level. Similarly, OGT disruption also leads to a parallel decrease in OGA protein levels (Kazemi et al., 2010). OGT knockdown cells by shRNA also displayed a reduced OGA level and decreased growth rate (Fig. S3A,B). These results reflect the observation that many cases of cancer tissues display increased protein levels of O-GlcNAc cycling enzymes, while aged tissues show decreased protein levels of O-GlcNAc cycling enzymes. Thus, we propose that aberrantly increased protein levels of O-GlcNAc cycling enzymes might be critical for cancer cell growth rather than total O-GlcNAcylation change.

In summary, this study provides key insight into the role of OGA in the maintenance of genomic instability linked to aging. OGA loss caused a reduced growth rate and cellular senescence with genomic instability. OGA enzymatic activity is important to complete mitosis with OGT functions. Abnormal OGA function might contribute to the aging process and be associated with development of age-related chronic diseases.

Materials and methods

Histology

For the histological analysis, mouse embryos, placentas, and tissues were collected, fixed with 10% phosphate-buffered formalin, dehydrated, and embedded in paraffin. The 6-μm sections were stained with hematoxylin and eosin by standard procedures. The slides were observed under a digital virtual microscope (dotSlide; Olympus, Tokyo, Japan).

Generation of MEFs and growth rate analysis

MEFs were isolated from 12.5 to 14.5 dpc embryos. After removal of the intestinal organs and head, embryos were washed with phosphate-buffered saline (PBS), minced, and trypsinized. After centrifugation, the dissociated cells were plated in Dulbeccos’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum. Growth curves were determined at passages 4 and 6. At day 0, 105 MEFs were seeded in triplicate on 6-cm plates, counted, and reseeded at the same density every 2–3 days. The population-doubling rate of each passage was calculated using a log2 formula (the number of harvested cells/the number of seeded cells).

Apoptosis assays using annexin-V and PI

OGA+/+ and OGA−/− MEFs were harvested, washed with ice-cold PBS, and resuspended in annexin-V binding buffer. Annexin-V–FITC (BD Biosciences, San Jose, CA, USA) and PI (Sigma, St. Louis, MO, USA) were added to the cells and incubated for 15 min at room temperature in the dark. Samples were analyzed by FACSCalibur (BD Biosciences).

Western blotting and immunofluorescence staining analysis

Tissue and cell lysates were prepared using standard procedures. Samples of 20–30 μg of protein were separated on an SDS-polyacrylamide gel and visualized. Western blotting antibodies used were anti-cyclin A (sc-596; Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-cyclin B1 (05-158; Upstate Biotechnology, Lake Placid, NY, USA), anti-cyclin E (sc-481; Santa Cruz Biotechnology), anti-cyclin D (sc-450; Santa Cruz Biotechnology), anti-AKB (611082; BD Biosciences), anti-cdc2 (9112; Cell Signaling Technology, Beverly, MA, USA), and β-actin (691001; MP Biomedicals, Santa Ana, CA, USA). The antibodies used for the anti-O-GlcNAc Western blots (CTD110.6) were kindly provided by Dr. Gerald Hart (Johns Hopkins University, Baltimore, MD, USA). Anti-OGT and OGA polyclonal antibodies had been previously generated and were used as previously described (Suh et al., 2008). Synthetic peptides (YAAGNKPDHMIK) for the anti-OGT antibody and purified His-OGA protein for the anti-OGA antibody were conjugated with keyhole limpet hemocyanin (KLH) (Calbiochem, San Diego, CA, USA) using glutaraldehyde. The peptide–KLH conjugates were immunized into a New Zealand white rabbit. The peroxidase-labeled goat anti-mouse IgG and the goat anti-mouse IgA, IgM, and IgG used as the secondary antibodies were obtained from Kirkegaard & Perry Laboratories (KPL, Gaithersburg, MD, USA).

For nuclei staining, the MEFs were grown on coverslips, fixed for 15 min at room temperature in 4% paraformaldehyde, and washed in PBS. DNA was visualized with Hoechst stain (Sigma), and the coverslips were mounted on slides. Nuclei were analyzed using a FV1000SPD (Olympus).

Cell cycle study

For primary MEF asynchronous cell cycle analysis, bromodeoxyuridine (BrdU) and PI staining was performed using BrdU flow kits (BD Pharmingen, San Diego, CA, USA). Briefly, the MEFs were incubated with 10 μm BrdU for 4 h. After harvesting and fixation, cells were permeabilized with BD Cytofix/Cytoperm™ buffer (BD Biosciences) and were labeled with an FITC-conjugated anti-BrdU antibody. Total DNA was stained with PI solution and analyzed using a FACSCalibur instrument (BD Biosciences).

Cytogenetic analyses

MEFs were treated with 0.1 μg mL−1 colcemid (Gibco/BRL, Grand Island, NY, USA) for 5 h. After washing with PBS, the cells were detached and incubated in 0.075 m KCl (hypotonic solution) at 37 °C for 15 min. The cells were fixed with methanol/glacial acetic acid (3:1) three times at room temperature for 10 min. The cells were dropped on a slide and stained with Giemsa.

Live cell imaging

Asynchronously grown MEFs were seeded on coverslips and maintained in a heated culture chamber (37 °C, 5% CO2) on Microscope equipped with a CO2 control system and a humidity chamber. The cells were visualized by DIC imaging. DIC images were collected at 10-min intervals for 24 h. Live images of cell division were obtained using an IX81-ZDC in UNIST-Olympus Biomed Imaging Center (UOBC).

Acknowledgments

This work was supported by the National Research Foundation of Korea Grant funded by the Korean Government (KRF-2007-341-C00027), a grant from the Korean Ministry of Education, Science and Technology (The Regional Core Research Program/Anti-aging and Well-being Research Center), the National Research Foundation of Korea Grant funded by the Korean government (no. 2010-0029434), and the Fusion Pioneer Project (PGB013) from the National Research Foundation of Korea.

Ancillary