Mammalian G9a is a euchromatic histone H3 lysine 9 (H3K9) methyltransferase essential for development. Here, we characterize the Drosophila homolog of G9a, dG9a. We generated a dG9a deletion allele by homologous recombination. Analysis of this allele revealed that, in contrast to recent findings, dG9a is not required for fly viability.
METHYLATION of several conserved lysine residues on histone H3 and H4 tails by histone methyltransferases (HMTases) plays a key role in chromatin structure and gene regulation. Methylation of H3K9 is generally associated with heterochromatin and gene repression and is a docking site for HP1 (Bannister et al. 2001; Jacobs et al. 2001; Lachner et al. 2001). Two H3K9-specific HMTases have been well characterized in Drosophila: SU(VAR)3-9 and DmSETDB1. SU(VAR)3-9 was shown to di- and trimethylate H3K9 at the chromocenter and is therefore considered as the heterochromatic H3K9 HMTase (Schotta et al. 2002; Ebert et al. 2004). DmSETDB1 mono- and dimethylates H3K9 in euchromatin, dimethylates H3K9 on chromosome 4, and is required for silencing of variegating transgenes on chromosome 4 (Seum et al. 2007). At least one additional HMTase is involved in H3K9 methylation in Drosophila, since H3K9 monomethylation at the chromocenter, H3K9 dimethylation at the telomeres, and some of the H3K9 mono- and dimethyl marks at euchromatic bands are not deposited by SU(VAR)3-9 nor DmSETDB1. One candidate for this function is the Drosophila homolog of the mammalian G9a, dG9a. Mammalian G9a is a euchromatic mono- and di-H3K9 methyltransferase (Tachibana et al. 2001; Rice et al. 2003). It is essential for early embryogenesis in mice and is involved in transcriptional silencing of developmentally regulated genes (Tachibana et al. 2002; Roopra et al. 2004; Feldman et al. 2006). Here, we generated a dG9a mutant that lacks the whole dG9a open reading frame (ORF) by homologous recombination. Analysis of this mutant and of flies overexpressing dG9a led us to conclude that this gene is not essential for fly viability and fertility and does not play a major role in H3K9 methylation in third instar larvae.
RESULTS AND DISCUSSION
CG2995 was identified by others and ourselves by blastp as the Drosophila homolog of mammalian G9a (Mis et al. 2006; Stabell et al. 2006). To characterize dG9a function, we first used the KG01242 line, where a P element is located 708 bp upstream of the dG9a ATG (Figure 1A). Because dG9a is expressed in this line (Figure 1B), it cannot be considered a dG9a mutant line. Therefore, we generated the dG9adel34 allele by imprecise excision of the P element, which removes most of the dG9a gene, except one ankyrin repeat and the SET domain (Figure 1A). The remaining part of the dG9a gene is transcribed (data not shown) and a potentially truncated dG9a protein can be produced. Moreover, the neighboring CG3038 gene is partially deleted. Therefore, we generated the dG9aRG5 deletion mutant by homologous recombination (Figure 1A). dG9aRG5 lacks the whole dG9a ORF. Both dG9aRG5 and dG9adel34 homozygotes are viable and fertile, with no particular phenotype. Thus, we conclude that dG9a is a nonessential gene and that flies lacking dG9a show no developmental defect. This contradicts results from others who concluded, on the basis of analysis of flies expressing dG9a RNAi, that dG9a is an essential gene involved in the ecdysone pathway and required for puparium formation (Stabell et al. 2006). We surmise that the phenotype observed upon dG9a RNAi expression (Stabell et al. 2006) is not related to the decrease in dG9a expression.
We next investigated whether dG9a is involved in H3K9 methylation. We analyzed the H3K9 mono-, di-, and trimethyl marks on third instar larvae polytene chromosomes of wild type and the dG9aRG5 homozygous mutant and could not show any difference between the two genetic backgrounds (supplemental Figure 1 at http://www.genetics.org/supplemental/). The H3K27 methylation pattern is also not altered in the absence of dG9a (supplemental Figure 2 at http://www.genetics.org/supplemental/). We also compared the H3K9 methylation levels in third instar larvae tissues (brain, salivary glands, imaginal discs) of wild-type and dG9aRG5 homozygotes by Western blot analysis and again could not detect any clear difference (Figure 2A and data not shown). Thus, dG9a is not responsible for H3K9 methylation in third instar larvae polytene chromosomes or tissues. But one could envisage that dG9a function is redundant with SU(VAR)3-9 or DmSETDB1. To address this point, we generated double homozygous mutants for dG9aRG5 and Su(var)3-906 and for dG9aRG5 and DmSetdb110.1a. Phenotypically, dG9aRG5–Su(var)3-906 double homozygotes behave as single Su(var)3-906 homozygotes, although the reduced viability observed for Su(var)3-906 homozygotes (Schotta et al. 2003) is more severe in the double mutant. As for the dG9aRG5–DmSetdb110.1a double mutant, it behaves as single DmSetdb110.1a homozygotes (Seum et al. 2007), as they die at late pupal stage. No additional phenotypic trait was observed in the double mutants when compared to Su(var)3-906 or DmSetdb110.1a single homozygotes. To address a potential redundancy in H3K9 HMTase activity between dG9a and SU(VAR)3-9 or DmSETDB1, we next analyzed the H3K9 mono-, di-, and trimethyl marks on third instar larvae polytene chromosomes in those genetic backgrounds (supplemental Figure 1). In dG9aRG5–Su(var)3-906 double homozygotes, the H3K9 methylation pattern is identical to that observed in Su(var)3-906 homozygotes, and, in dG9aRG5–DmSetdb110.1a double homozygotes, the H3K9 methylation pattern is identical to that of DmSetdb110.1a homozygotes (see supplemental Figure 1, A–C, and its legend for detailed comments on the H3K9 methylation patterns). Thus we conclude that dG9a displays no H3K9 methyltransferase activity in third instar salivary glands and is not redundant with respect to the other H3K9 methyltransferases SU(VAR)3-9 and DmSETDB1. Still, we cannot exclude that dG9a methylates H3K9 in a temporally or spatially restricted fashion. In any event, if this were truly the case, then this dG9a activity would not be required for viability, as dG9aRG5 homozygotes show no phenotype.
To further characterize a potential HMTase activity of dG9a, we overexpressed a HA-tagged version of dG9a. For this purpose, HA-dG9a was expressed under the control of the hsp70 promoter and induced by three heat shocks of 30 min at 37°/day during the entire period of development. Overexpression of HA-dG9a does not interfere with fly viability. On salivary gland polytene chromosomes, HA-dG9a localizes at the euchromatic arms and is not present at the chromocenter or at chromosome 4 (Figure 2B), as described for the endogenous dG9a (Stabell et al. 2006). Staining of polytene chromosomes of overexpressing HA-dG9a larvae with antibodies recognizing H3K9 mono-, di-, and trimethyl and H3K27 mono-, di-, and trimethyl did not reveal an increase in any of those six marks (data not shown). We also compared the H3K9 methylation levels between wild-type flies and flies overexpressing HA-dG9a in third instar larvae tissue and could detect a slight increase of H3K9 dimethyl upon HA-dG9a overexpression (Figure 2A); as a positive control, overexpression of 3HA-DmSETDB1 leads to a more evident increase of H3K9 dimethylation (Figure 2A). H3K9 mono- or trimethyl levels are not influenced by the overexpression of dG9a (data not shown). This argues in favor of a role for dG9a in H3K9 methylation in Drosophila. This is consistent with in vitro HMTase assays performed by others, who showed that truncated dG9a can dimethylate H3K9 (Mis et al. 2006; Stabell et al. 2006). In vitro, dG9a was also shown to mono- and trimethylate H3K9 and to methylate H3K27, as well as H4, at K8, K12, or K16 (Mis et al. 2006; Stabell et al. 2006). However, this needs to be confirmed in vivo with the full-length protein, as we could not detect any H3K9 mono- or tri- or H3K27 methyltransferase activity of dG9a in vivo.
Taken together, these results show that dG9a is not essential for viability and fertility and that the absence or overexpression of dG9a does not lead to any change in the H3K9 methylation pattern on third instar larvae polytene chromosomes. In addition, this work shows that dG9a is not redundant with respect to the H3K9 methyltransferases SU(VAR)3-9 and DmSETDB1. However, upon HA-dG9a overexpression, a slight increase in H3K9 dimethylation in third instar larvae tissues suggests a potential function of dG9a in H3K9 methylation that has still to be fully characterized at the spatial, temporal, and functional levels.
We thank Susan Opravil and Thomas Jenuwein for generous gifts of antibodies and Bernard Conrad for comments on the manuscript. This work was supported by the Swiss National Science Foundation and the State of Geneva.
Communicating editor: J. A. Birchler
- Received June 29, 2007.
- Accepted August 28, 2007.
- Copyright © 2007 by the Genetics Society of America