(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Dynamic evolution of the heterochromatin sensing histone demethylase IBM1 [1]

['Yinwen Zhang', 'Department Of Genetics', 'University Of Georgia', 'Athens', 'Georgia', 'United States Of America', 'Hosung Jang', 'Ziliang Luo', 'Yinxin Dong', 'Yangyang Xu']

Date: 2024-07

Heterochromatin is critical for maintaining genome stability, especially in flowering plants, where it relies on a feedback loop involving the H3K9 methyltransferase, KRYPTONITE (KYP), and the DNA methyltransferase CHROMOMETHYLASE3 (CMT3). The H3K9 demethylase INCREASED IN BONSAI METHYLATION 1 (IBM1) counteracts the detrimental consequences of KYP-CMT3 activity in transcribed genes. IBM1 expression in Arabidopsis is uniquely regulated by methylation of the 7th intron, allowing it to monitor global H3K9me2 levels. We show the methylated intron is prevalent across flowering plants and its underlying sequence exhibits dynamic evolution. We also find extensive genetic and expression variations in KYP, CMT3, and IBM1 across flowering plants. We identify Arabidopsis accessions resembling weak ibm1 mutants and Brassicaceae species with reduced IBM1 expression or deletions. Evolution towards reduced IBM1 activity in some flowering plants could explain the frequent natural occurrence of diminished or lost CMT3 activity and loss of gene body DNA methylation, as cmt3 mutants in A. thaliana mitigate the deleterious effects of IBM1.

In flowering plants, IBM1 histone demethylase plays a crucial role in regulating chromatin structure by removing H3K9me2, a modification associated with heterochromatin. This process involves a distinctive mechanism where the methylation of a repetitive sequence within an intron allows IBM1 to monitor and respond to global H3K9me2 levels. We discovered that this intron methylation sensor is widely observed across flowering plants, although the exact sequences involved show considerable variation. This suggests that while the mechanism for controlling IBM1 activity is conserved, it has adapted differently in various plant species. Additionally, our findings include the identification of Arabidopsis thaliana accessions that mimic weak ibm1 mutants and several Brassicaceae species with diminished IBM1 expression or deletions, which correlate with reduced CMT3 activity and gene body DNA methylation.

This study explores the association between intron DNA methylation of IBM1 and its role in its own expression by surveying within and between species variation. A. thaliana accessions were identified that were reminiscent of weak A. thaliana ibm1 mutants, as they possessed ectopic mCHG in a subset of genes. Furthermore, a comparative analysis of IBM1 orthologs across 34 angiosperm species demonstrated the presence of intronic DNA methylation within its 7th intron, indicating the evolutionary conservation of the H3K9me2 sensor in flowering plants. However, the sequence underlying the methylated intron was highly variable between species suggesting this heterochromatin sensing activity exhibits significant evolutionary divergence. Moreover, our investigation into multiple Brassicaceae species suggests the coevolution of IBM1 and CMT3 within this family and likely all flowering plants. This is particularly evident in Brassicaceae species that lack gbM, such as Eutrema salsugineum and Thlaspi arvense, as we observed a correlation between low or absent CMT3 expression and reduced IBM1 expression. This association was further supported by DNA methylome data from other Brassicaceae species that have reduced/absent IBM1 and/or CMT3 function as well as gene body DNA methylation. Collectively, our study shows that IBM1, its intronic heterochromatin sensor and CMT3 are dynamically evolving and that this shapes the genic methylation landscape in plants.

A unique aspect of IBM1 is its dependency on DNA and H3K9 methylation within its large 7th intron for transcriptional and post-transcriptional regulation [ 40 – 42 ]. IBM1 is a ubiquitously expressed gene and is known to produce two distinct mRNA isoforms in A. thaliana. IBM1-L, the longer isoform, encodes a functional protein with a catalytic JmjC histone demethylase domain, whereas its shorter counterpart, IBM1-S, is non-functional without the catalytic JmjC domain [ 40 ]. Notably, the expression of these isoforms is influenced by DNA methylation within the IBM1 intron. In the case of A. thaliana Col-0, the 7th intron of IBM1 contains DNA methylation, crucial for the expression of the functional IBM1-L isoform [ 40 – 42 ]. Full-length IBM1 has the capability to remove H3K9me2 in genic regions, suggesting that intron methylation in IBM1 serves as a regulatory sensor of H3K9me2 by modulating the balance of its transcript isoforms [ 33 , 34 , 40 ]. Furthermore, our previous study revealed that certain natural A. thaliana accessions exhibit increased mCHG in genic regions, some of which also show decreased intron methylation in IBM1, along with an alteration in the IBM1-S/IBM1-L ratio compared to Col-0 [ 26 ]. This suggests a potential interaction between IBM1 expression level and intron methylation level in A. thaliana accessions, and raises questions about the extent to which this intron methylation sensor mechanism is conserved among different A. thaliana accessions and across other flowering plant species in shaping the epigenome.

The binding activity of KYP is not limited to mCHG, as KYP also engages with mCG prominently present in a specific group of genes classified as gene body methylated (gbM) [ 15 , 17 – 20 ]. These gbM genes typically include ‘housekeeping’ genes with moderate expression, characterized by extended gene lengths, lower substitution rates (dN/dS), a higher prevalence of CWG (W = A or T, cytosines preferred by CMT3), and fewer CG dinucleotides [ 21 – 26 ]. Despite the ongoing debate about the role of gbM in plants, one study suggests its role could be to suppress antisense transcripts within a subset of gbM regions [ 27 ]. One prevailing hypothesis is that CMT3 is important for the initial establishment of gbM, primarily through influencing CHG methylation [ 28 – 31 ]. This activity is thought to subsequently facilitate CG methylation in the gene body. Once established, CG methylation is maintained by MET1, ensuring the stability of methylation patterns across cell divisions and subsequent generations [ 6 ]. The observed natural absence of CMT3 in some angiosperm species correlates with a loss of gbM and also highlights its role in a maintenance phase [ 28 , 31 ]. MET1 is the primary contributor to the maintenance of CG methylation within gene bodies, which is generally unaffected in Arabidopsis cmt3 mutants [ 28 ]. However, genetic evidence supports its function in de novo mCG as well [ 32 ]. The presence of mCG within gbM genes likely facilitates KYP binding, recruiting the CMT3-KYP heterochromatin complex and exposing these genes to silencing machinery. However, the CMT3-KYP heterochromatin feedback loop in genic regions is disrupted by the histone lysine demethylase, INCREASED IN BONSAI METHYLATION1 (IBM1), which selectively demethylates H3K9me2 in genes, thus safeguarding them from silencing [ 33 – 36 ]. This protective role of IBM1 is underscored in ibm1 mutants, which exhibit diverse phenotypic abnormalities and an accumulation of H3K9me2 and mCHG in approximately one-fifth of coding genes [ 33 , 34 , 36 – 38 ]. These affected genes in ibm1 predominantly belong to the category of gbM genes [ 39 ], indicating a targeted recruitment of the CMT3/KYP complex to these specific loci. The dynamic interplay between IBM1 and CMT3/KYP is important for maintaining the equilibrium between euchromatin and heterochromatin, suggesting a co-evolutionary relationship [ 39 ]. In Arabidopsis thaliana, the seed fertility defect and meiotic abnormalities observed in ibm1 is rescued by knocking out CMT3, indicating a functional interdependence [ 36 , 38 ]. Furthermore, the exclusive presence of both IBM1 and CMT3 in flowering plants supports the evolutionary connection between these two genes [ 30 ].

DNA methylation and histone H3 lysine 9 (H3K9) methylation are essential repressive chromatin modifications required for the formation of heterochromatin and the silencing of transposable elements (TEs), thereby playing a key role in maintaining genomic stability [ 1 – 5 ]. In plants, DNA methylation is observed in three different contexts: CG, CHG, and CHH (where H—A, C or T), each maintained by specific DNA methyltransferases. METHYLTRANSFERASE1 (MET1) is responsible for sustaining CG methylation through DNA replication [ 6 ], CHROMOMETHYLASE3 (CMT3) facilitates CHG methylation working in concert with the H3K9 methyltransferase KRYPTONITE (KYP) (also called SUVH4) [ 6 – 10 ], whereas CHH methylation is established either through the activities of CMT2 or the RNA-directed DNA methylation (RdDM) pathway [ 10 , 11 ]. All three contexts of DNA methylation are predominantly localized in heterochromatin and TE/repeat regions where CHG methylation (mCHG) is particularly important for reinforcing heterochromatin DNA methylation in conjunction with H3K9me2 [ 7 , 8 , 12 – 17 ]. This synergy is largely due to the unique characteristics of the enzymes CMT3 and KYP, as CMT3 preferentially binds to H3K9me2, and uses it as a guide to deposit mCHG and KYP, which recognizes pre-existing DNA methylation, adds H3K9me2 [ 8 , 15 – 17 ]. This interplay between CMT3 and KYP establishes a positive feedback mechanism, reinforcing the accumulation of both mCHG and H3K9me2 within heterochromatin regions [ 14 , 16 ].

Results

The ectopic mCHG in genes in Cnt-1 is reduced by IBM1 overexpression The A. thaliana Cnt-1 accession stands out as an exceptional case, exhibiting the largest number of ectopic mCHG-gain genes among the natural accessions along with lower levels of intron methylation (Fig 1A). As this phenotype resembles that of a weak ibm1 mutant, we explored isoform quantification analysis of Cnt-1 to estimate IBM1’s expression level. Notably, Cnt-1 displayed a substantially lower expression level of the functional IBM1-L isoform in comparison to the Col-0 reference accession, with only a marginal reduction in the expression of the IBM1-S isoform (Fig 2A). This result suggests that decreased expression of IBM1-L in Cnt-1 leads to an elevation in mCHG within these genes. Moreover, TE methylation in Cnt-1 is relatively high among all accessions (S1M Fig and S1 Table). This combination of subdued IBM1 activity and vigorous CMT-KYP methylation may account for the high number of mCHG-gain genes in Cnt-1 and supports that this heterochromatin sensing mechanism is dynamic in populations to shape the epigenome. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. The ectopic mCHG in genes in Cnt-1 is reduced by IBM1 overexpression. (A) Expression level of IBM1-L and IBM1-S in Col-0 and Cnt-1. (B) Expression level of IBM1 in each transgenic plant. Primers were designed to target the conserved region of IBM1 between Col-0 and Cnt-1. RT-qPCR was performed to estimate the expression level of IBM1 in these plants. (C) A genome browser view shows that ectopic mCHG in genes in Cnt-1 is reduced by introducing the IBM1 transgene. (D) Metaplots show the average changes to mCHG over gbM and UM genes in IBM1 transgenic lines of Cnt-1. (E) The number of mCHG-gain genes is reduced in IBM1 transgenic lines compared to Cnt-1. (F) Heatmaps show reduction of mCHG on genes in IBM1 transgenic lines of Cnt-1. The 636 mCHG-gain genes from (E) are shown. https://doi.org/10.1371/journal.pgen.1011358.g002 Next, we examined the effect of restoring IBM1 expression in the Cnt-1 accession. The UBQ10 cis-regulatory sequences were used to express the Col-0 IBM1 coding sequence in Cnt-1, resulting in the generation of three independent transgenic lines. These lines exhibited a significant increase in IBM1 expression, ranging from 20-100-fold higher than the control (Fig 2B). All three of the IBM1-ox lines displayed a decrease in mCHG in a subset of gene bodies compared to the Cnt-1 control (Fig 2C and 2D), indicating that overexpression of IBM1 reduces ectopic mCHG in gene bodies presumably by decreasing H3K9me2, although this was not tested. In Arabidopsis, H3K9me2 and mCHG form a self-reinforced loop, thus, measuring mCHG levels can indirectly reflect H3K9me2 levels [8,16,17]. This suggests that the observed reduction in mCHG are likely indicative of decreased H3K9me2 levels. This reduction was further supported by a decrease in the number of mCHG-gain genes when compared to the Cnt-1 control (Fig 2E). A heatmap analysis of mCHG-gain genes from Cnt-1 revealed a substantial decrease in mCHG levels in the IBM1-ox lines (Fig 2F). Despite the substantial increase in expression levels (Fig 2B), complete mCHG reduction was not achieved (Fig 2C–2F), which could be attributed to promoter choice. Although we used the UBQ10 promoter for its strong and constitutive expression, this may not fully replicate the native expression pattern of IBM1, particularly in tissue-specific contexts where IBM1 function is crucial, such as during meiosis [38]. This discrepancy could explain the partial reduction in mCHG levels observed, suggesting that the expression driven by UBQ10 does not fully mimic the native promoter’s regulatory effects. Considering the conservation of IBM1 coding sequences between Col-0 and Cnt-1 (99.61% sequence identity), these findings collectively indicate that increasing IBM1 expression in Cnt-1 significantly reduces the aberrant mCHG accumulation in the gbM genes. We extended the analysis of methylation beyond gene regions to include TEs and TE genes (S3A Fig). When comparing the methylation profiles of the Col-0 and Cnt-1 lines, both generally exhibited similar methylation levels. However, the IBM1-ox lines displayed methylation patterns similar to Cnt-1 for mCG, while mCHG and mCHH levels were slightly reduced. Furthermore, we assessed global DNA methylation levels (S3B Fig), revealing that, aside from minor differences in mCG levels compared to the Cnt-1 control, there were notable reductions in mCHG and mCHH levels across the pericentromeric regions of the genome. This suggests that IBM1 overexpression has a broad impact on methylation, affecting regions well beyond gene bodies.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011358

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.