Decoding G9a histone methyltransferase: The role of G9a in H3K9 methylation and regulation of its own alternative splicing
Chromatin modifications are closely linked to alternative splicing with functional consequences for varying differentiation programs. G9a is a catalytic SET domain-containing protein that localizes to euchromatic histone for the methylation of H3 lysine 9 and it exists as two isoforms through the differential splicing of exon 10 (E10). Euchromatic H3K9 methylation, which has been linked to the HMTase domain of G9a, leads to suppressed transcription levels of developmental genes such as Mage-a. Fiszbein et al. demonstrate that G9a HMTase activity on its chromatin indirectly enhances E10 inclusion in its transcript, thereby regulating alternative splicing of its own transcript. By regulating its own alternative splicing, G9a promotes neuron differentiation and creates a positive feedback loop that bolsters cellular commitment to differentiation.
Dynamic chromatin structure allows the eukaryotes to regulate DNA processes in response to various stimuli and circumstances. As a result, factors the affect the formation of chromatin, like acetylases and methylases, have significant consequences for understanding and manipulating specific DNA processes. Histone methylation has been linked to a diversity of functions such as heterochromatin formation, X-chromosome inactivation, and transcriptional regulation. Due to their crucial role in epigenetic regulation, enzymes involved in histone modifications have recently gained scientific interest to understand diseases such as fetal alcohol spectrum disorders. Recent identification of several histone methyltransferases (HMTases) presents opportunities to refine our understanding of the biological implications of these chromatin modifications and potentially develop therapeutics in response.
In mammalian HeLa cells, lysines 4, 9, and 27 of H3 and lysine 20 in H4 are methylated (Tachibana et al., 2001). Suv39 was the first identified lysine-preferring mammalian HMTase that acts upon lysine 9 of H3. Suv39 family proteins are part of the SET domain-containing protein families and also possess a chromo-domain, a chromatin regulator motif. In light of this discovery, G9a, another SET domain-containing protein with a chromo-domain, has been previously identified as another candidate for H3K9 methylation (Tachibana et al., 2001). Tachibana et al. found nuclear localization of G9a to be different from Suv39, which primarily localizes to repressive chromatin domains, suggesting that G9a has a biologically distinct role from that of Suv39.
In the first highlighted experiment, Tachibana et al. investigate the in vivo functions of G9a in G9a-deficient mice and embryonic stem (ES) cells and the effect of its HMTase activity on transcriptional regulation. In the second experiment, Fiszbein et al. asks how the methyltransferase activity of G9a affects the differential splicing of exon 10. They look at how different splice variants of E10 influences neuronal differentiation, H3K9 methylation levels, and localization patterns of G9a protein. Curiously, their data suggest that G9a self-regulates its own alternative splicing through a positive feedback loop. Taken together, these experiments ask how chromatin structure can regulate alternative splicing and how splicing variation contributes to functional diversity to serve a greater biological purpose.
G9a exerts a transcriptionally suppressive function dependent on its HMTase activity
(Tachibana et al., 2002)
In a 2001 study, Tachibana et al. found evidence to believe that G9a, a mammalian HMTase, methylates H3K9 in non-heterochromatic loci. Their previous experiments distinguished G9a HMTase activity toward H3-K9 from Suv39h1. Data in their previous study demonstrate the nuclear localization of G9a to transcriptionally active euchromatin compared to the localization of Suv39 h1 to repressive chromatin domains of centromeric loci. In these experiments, Tachibana et al. found that G9a deficiency resulted in lethality or severe growth and development defects in the embryos. After determining that G9a function was crucial for differentiated somatic cells, the researchers found that G9a deficient embryos also demonstrated drastically decreased levels of H3K9 methylation and that the defect was completely rescued with the expression of exogenous G9a protein (Figure 3). The authors determined that G9a is the major in vivo H3K9 HMTase that directs methylation of euchromatic regions. In light of these data, the authors asked the function of G9a-mediaated histone methylation in the regulation of transcription.
In order to do this, Tachibana et al. used a luciferase reporter gene assay in order to discern the role of G9a-mediated histone methylation in transcription. The reporter vector contained the firefly luciferase gene, human DNA polymerase ? promoter, five GAL4-binding sites, and an SV40 enhancer (Sekimata et al., 2001). In additional to measuring the expression of the gene-of-interest, reporter gene assays can study transcriptional regulation by assessing the expression of the reporter gene driven by the clone promoter fragment of the gene-of-interest (Solberg, N. & Krauss, S., 2006). In this experiment, the pol ? promoter fused to G9a gene was used to drive the expression of the luciferase gene. The reaction that luciferase catalyzes was quantified as luminescence, which can then be measured to reflect gene expression.
The expression constructs for the plasmids used for the luciferase assays contained the GAL4 DNA-binding domain fused with one of the five different domains (Figure 5A). The PM-G9a-L construct expressed the full-length mG9a-L. The PM-HMT expressed just the HMTase domain of G9a. PM-G9a-S?NHLC expressed the full-length mG9a-S with a deletion in the HMTase domain and the PM-?NHLC construct expressed the HMTase domain with the mutation alone. The PM construct expressed the GAL4-DBD alone.
Protein expression of the constructs shown in Figure 5A was then detected using a Western blot and a ?-GAL4-DBD antibody (Figure 5B). We see a band at approximately 20 kDA in each lane, which corresponds with a band at 20 kDA for the PM construct only containing the GAL4-DBD. This ensures that the observations of the reporter gene assay were not due to a lack of protein. Luminescence was then measured and quantified for each of the five constructs as shown in Figure 5C. The PM construct is used as a negative control since it only contains the GAL4-DBD fused to the reporter gene and expressed the reporter gene constitutively. If G9a was implicated in transcriptional regulation, the results should have indicated a change in luminescence for the PM-G9a-L and the PM-HMT constructs, which is what the authors observed. The luciferase assay showed decreased levels of luminescence for the PM-G9a-L and PM-HMT constructs, both of which expressed the HMTase domain. Levels of luminescence comparable to the control are shown for the PM-?NHLC and PM-G9a-S?NHLC constructs, which contained the mutated HMTase domains (Fig. 5B). These results indicated that G9a had a transcriptionally suppressive effect that depended on the expression of the HMTase domain of the G9a gene.
In order to elucidate the mechanism by which G9a regulates transcription, Tachibana et al. performed the luciferase assay in the presence of trichostatin A (TSA), an inhibitor of histone deacetylases (HDACs) to see if levels of transcription could be rescued with TSA. Since histone acetylation is correlated with transcriptional activation, HDACs allow histones to bind more tightly to DNA and suppress transcription. Thus, transcriptional suppression attributed to HDAC activity should be relieved in the presence of TSA. Although acetylation of the histones significantly increased in the presence of TSA (upper right panel of Fig. 5D), levels of luminescence of the PM-G9a-L and PM-HMT constructs remained decreased compared to the control. Decreased levels of transcription by these constructs even upon TSA treatment suggests that transcriptional regulation does not occur through the same mechanisms of the HDAC pathway. These results further corroborate the results in Figure 5B, suggesting that G9a HMTase-mediated repression of gene transcription occurs in a manner independent of the HDAC pathway.
Data from the reporter gene assay are reinforced by studies of Mage-a—the transcription of which is regulated by G9a. Using chromatin immunoprecipitation, the authors found increased levels of H3K9 methylation in chromatin containing the Mage-a2 promoter sequences in the mouse embryonic stem (ES) cells homozygous wild-type for G9a and decreased levels in ES cells homozygous mutant for G9a. They also found that expression of exogenous G9a in mutant G9a ES cells rescued H3K9 methylation levels to wild-type G9a ES cell levels. These data further strengthen the conclusions drawn from Figure 5. Overall, these data strongly contend that euchromatic H3K9 methylation, regulated by G9a, is transcriptionally repressive and is dependent on its HMTase activity.
Figure 2. G9a Methyltransferase Activity is Involved in the Regulation of Its Own Alternative Splicing (Fiszbein, et al., 2016)
The previously highlighted experiment provides data for G9a as the methyltransferase responsible for euchromatic H3K9 methylation. In this study, Fiszbein et al. (2016) characterized the role of G9a in the regulation of its alternative splicing of exon 10 (E10). G9a exists as two isoforms with differential inclusion of E10 through alternative splicing. The authors studied alternative splicing of E10 due to its observed upregulation in mature neurons. They found that E10 inclusion in G9a is required for neuron differentiation and that H3K9 methylation is stimulated during neuron differentiation independent of intrinsic G9a HMTase activity. These data suggest that E10 inclusion increases methylation efficiency by facilitating nuclear import of G9a, indicated by E10-promoted G9a nuclear localization (Figure 6). Since the upregulation of E10 inclusion correlates with increased H3K9me2 levels in G9a gene, the authors analyzed if chromatin structure played a role in the regulation of G9a alternative splicing.
In other experiments described in the paper, the authors determined the promotion of nuclear localization of G9a by E10 inclusion and increased H3K9me2 levels during neuronal differentiation (Figures 4, 5, and 6). Thus, they hypothesized that the inclusion of E10 would promote nuclear localization, increase methylation of the G9a gene, and these chromatin modifications would lead to different alternative splicing patterns. To test this, the researchers used a minigene assay and RT-PCR. The minigenes were designed to include exons 9 to 11 and their respective introns, had the same splicing transition as the endogenous gene (Figure S5B), and were transfected into HeLa and HEK293 cells, which show “non-constitutive” splicing. These data were expressed as a ratio of mRNAs including and excluding exon 10 (E10+/ E10-) (Cooper, 2005). One-way or two-way ANOVA statistic analysis with a Bonferroni correct was used to reduce Type I error (“false positive” reading) because the researchers are testing multiple hypotheses under different conditions.
To begin, they looked at the effect of G9a methyltransferase activity on E10 inclusion induced by cell differentiation. They treated differentiated cells with either vehicle or BIX-01294, a selective inhibitor of G9a methyltransferase and H3K9 methylation. If G9a is involved in its own splicing of E10, then the inhibition of its activity via BIX should change E10+/E10- ratios. RT-PCR gels revealed high levels of E10 inclusion in cells differentiated with RA after three days that significantly decreased upon treatment with the BIX inhibitor (Figure 7A). In contrast, the data indicates an enhancing effect in undifferentiated cells treated with BIX compared to the vehicle, which was used as a negative control. Since BIX had a suppressive effect on differentiation-dependent E10 inclusion, one might conclude that G9a activity is somehow implicated in the process.
To further deduce the role of G9a activity in the alternative splicing of E10, the researchers performed a co-transfection of the G9a splicing reporter minigene and expression constructs encoding G9a isoforms, E10+, E10- full-length cDNAs, and empty vectors in differentiated N2 cells. Gels demonstrate high levels of E10 in cells transfected with the E10+ construct rather than the E10- and empty vectors (Figure 7B). The experiment was also performed without the reporter minigene (- reporter vector) as a negative control. These data demonstrate that the inclusion of E10 precipitates more E10+ isoforms than the E10- isoform and the empty vector.
Subsequently, the authors mutated the catalytic site of G9a and saw the effects of the mutation on G9a alternative splicing. They transfected an E10 + isoform with a dead catalytic site into HeLa cells and G9a CRISPR-ablated HEK293 cells to test whether or not the presence of the E10 isoform was sufficient for alternative splicing. RT-PCR analysis revealed significantly decreased E10 inclusion in the catalytically dead mutant (G9a E10+ mutant) compared to the wild-type (G9aE10+ wt) (Figure 7C). A stronger inhibitory effect was also seen when two types of catalytic mutants (G9a E10+mu1 and G9a E10+ mu2) were transfected into G9a knockout cells (Figure 7D). In these data, we can see that G9a minigene E10+ expression was almost completely suppressed with the G9a E10+ mut2, in which the G9a-GLP interaction was undermined when compared to mut1. These data, taken together, suggest that protein-protein interactions between G9a and GLP, a mammalian lysine methyltransferase thought to interact with G9a in a heteromeric complex, is required for nuclear import, in addition to G9a catalytic activity in the regulation of E10 splicing variation (Tachibana, 2008).
The researchers then treated cells with the empty vector, the G9a E10- construct, or the G9a E10+ vector with either BIX or vehicle. If G9a is involved in its own alternative splicing, as we will see, we should see a change in the pattern of E10 inclusion when G9a cannot perform its function. They saw the most significant decrease in E10+ inclusion for G0a E10+ construct and a negligible difference for the empty and G9a E10- constructs (Figure 7E). A FRET reporter system measured interactions between the K9 methylation site of histone H3 and the methyllysine binding domain and revealed that E10 inclusion has no effect on G9a HMTase activity, suggesting that increased H3K9 methylation by E10 inclusion is achieved by way other than G9a activity (Figure 4). These data, understood with respect to previous experiment highlighted in Figure 4, suggest that E10 is really controlling nuclear import rather than enhancing the HMTase domain to increase H3K9 methylation levels and alternative splicing. Therefore, BIX treatment should not affect E10- mutant because G9a E10- mutant proteins remain in the cytoplasm, unable to access the nucleus. The authors show differential inclusion of E10 does not affect the inherent catalytic activity but does affect the methylation efficiency of histones by regulating access to the nucleus (Figures 4, 7C, 7D, and 7E).
The researchers proposed that G9a localizes to the nucleus in the event of differentiation—an event that favors the E10+ G9a alternative splicing pattern. This event precipitates increased H3K9 methylation of the G9a gene, which requires the activity of the G9a methyltransferase protein (Figure 7B). They dismissed the hypothesis that this may be due to increased G9a intrinsic catalytic activity because they found that E10+ splicing pattern does not affect G9a catalytic activity in other studies (Figure 4). The studies conducted by Fiszbein et al. suggest G9a influences the regulation of its own alternative splicing of E10—the inclusion of which regulates access of the protein to the nucleus and thus the activity of G9a.
Tachibana et al. (2002) demonstrate G9a’s transcriptional repressive activity as a function of its histone methyltransferase activity on chromatin structure. These changes in chromatin structure have consequences for regulation of its own alternative splicing through a positive feedback loop and its implications for cell differentiation. In the positive feedback loop, inclusion of E10 in G9a is required for neuron differentiation and neuronal differentiation triggers nuclear localization of G9a. E10 inclusion stimulates H3K9 methylation of the G9a gene, which requires the HMTase function of the G9a protein. Changes in chromatin structure, in turn, affect the regulation of G9a alternative splicing, which affects neuron differentiation and feeds into the positive feedback loop (Fiszbein et al., 2016).
The Fiszbein et al. paper links alternative splicing of G9a to the regulation of neuron cell differentiation—one example of how changes on a molecular scale translates to larger biological significance. G9a has been found to be a prominent character in many biological processes such as cell differentiation, genomic imprinting, and chromosome stability (Zhang et al., 2016). G9a has also been found to regulate cell differentiation through other mechanisms such as methylation of transcription of transcription factors, as is the case with skeletal muscle cell differentiation (Tze Ling, et al., 2011). Understanding the role of G9a also holds clinical potential in treating and developing therapeutics as well. Studies conducted on neonatal mouse brain indicate that G9a methyltransferase activity plays a critical role in regulating ethanol-induced neurodegeneration in the developing brain, which has potential consequences for understanding fetal alcohol spectrum disorders.
Alternative splicing optimizes the generation of multiple, diverse mRNAs from a limited numbers of genes, expanding the coding capacities of genomes. Differences in alternative splicing profiles contribute to isoform expression and influence cell differentiation and cell fate determination. Developing studies reveal the importance of epigenetic regulators and alternative splicing on directing specific transcriptional programs in neuron differentiation. G9a regulating its own alternative splicing to reinforce neuron differentiation is one biological snapshot of how alternative splicing manifests into physiological functions that arise in different developmental processes in humans and animals.