Covalent modification of histone tails is crucial for transcriptional regulation, mitotic chromosomal condensation, and heterochromatin formation. Histone H3 lysine 9 (H3-K9) methylation catalyzed by the Suv39h family proteins is essential for establishing the architecture of pericentric heterochromatin. We recently identified a mammalian histone methyltransferase (HMTase), G9a, which has strong HMTase activity towards H3-K9 in vitro. To investigate the in vivo functions of G9a, we generated G9a-deficient mice and embryonic stem (ES) cells. We found that H3-K9 methylation was drastically decreased in G9a-deficient embryos, which displayed severe growth retardation and early lethality. G9a-deficient ES cells also exhibited reduced H3-K9 methylation compared to wild-type cells, indicating that G9a is a dominant H3-K9 HMTase in vivo. Importantly, the loss of G9a abolished methylated H3-K9 mostly in euchromatic regions. Finally, G9a exerted a transcriptionally suppressive function that depended on its HMTase activity. Our results indicate that euchromatic H3-K9 methylation regulated by G9a is essential for early embryogenesis and is involved in the transcriptional repression of developmental genes.

The covalent modification of histone tails has regulatory roles in various nuclear processes, such as control of transcription and mitotic chromosome condensation. Among the different groups of enzymes known to catalyze the covalent modification, the most recent additions are the histone methyltransferases (HMTases), whose functions are now being characterized. Here we show that a SET domain-containing protein, G9a, is a novel mammalian lysine-preferring HMTase. Like Suv39 h1, the first identified lysine-preferring mammalian HMTase, G9a transfers methyl groups to the lysine residues of histone H3, but with a 10 -20-fold higher activity. It was reported that lysines 4, 9, and 27 in H3 are methylated in mammalian cells. G9a was able to add methyl groups to lysine 27 as well as 9 in H3, compared with Suv39 h1, which was only able to methylate lysine 9. Our data clearly demonstrated that G9a has an enzymatic nature distinct from Suv39 h1 and its homologue h2. Finally, fluorescent protein-labeled G9a was shown to be localized in the nucleus but not in the repressive chromatin domains of centromeric loci, in which Suv39 h1 family proteins were localized. This finding indicates that G9a may contribute to the organization of the higher order chromatin structure of non-centromeric loci.

During apoptotic execution, chromatin undergoes a phase change from a heterogeneous, genetically active network to an inert highly condensed form that is fragmented and packaged into apoptotic bodies. We have previously used a cell-free system to examine the roles of caspases or other proteases in apoptotic chromatin condensation and nuclear disassembly. But so far, the role of DNase activity or ATP hydrolysis in this system has not yet been elucidated. Here, in order to better define the stages of nuclear disassembly in apoptosis, we have characterized the apoptotic condensation using a cellfree system and time-lapse imaging. We demonstrated that the population of nuclei undergoing apoptosis in vitro appears to follow a reproducible program of nuclear condensation, suggesting the existence of an ordered biochemical pathway. This enabled us to define three stages of apoptotic chromatin condensation: Stage 1 ring condensation; Stage 2 necklace condensation; and Stage 3 nuclear collapse/disassembly. Electron microscopy revealed that neither chromatin nor detectible subnuclear structures were present inside the stage 1 ring-condensed structures. DNase activity was not essential for stage 1 ring condensation, which could occur in apoptotic extracts depleted of all detectible DNase activity. However, DNase(s) were required for stage 2 necklace condensation. Finally, we demonstrated that hydrolysable ATP is required for stage 3 nuclear collapse/disassembly. This requirement for ATP hydrolysis further distinguished stage 2 from stage 3. Together, these experiments provide the first steps towards a systematic biochemical characterization of chromatin condensation during apoptosis.

No abstract

The human serine/threonine kinase Aurora-B is structurally related to the protein kinase Ipl1p from S cerevisiae and aurora from Drosophila melanogaster, which are key regulators of mitosis. The present study shows that human Aurora-B is activated by okadaic acid and forms complexes with the protein serine/threonine phosphatase type 1 (PP1) or PP2A, but not with PP5. These data identi®ed Aurora-B associated protein phosphatases as negative regulators of kinase activation. We then used a series of substrates based on a histone H3 phosphorylation site (residues 5 ± 15) to determine the substrate speci®city of human Aurora-B. We found that this enzyme is an arginine-directed kinase that can phosphorylate histone H3 at serines 10 and 28 in vitro, suggesting that human Aurora-B is a mitotic histone H3 kinase.