(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:
https://journals.plos.org/plosone/s/licenses-and-copyright
------------
RNA-directed DNA methylation prevents rapid and heritable reversal of transposon silencing under heat stress in Zea mays
['Wei Guo', 'Department Of Botany', 'Plant Pathology', 'Purdue University', 'West Lafayette', 'Indiana', 'United States Of America', 'Dafang Wang', 'Division Of Math', 'Sciences']
Date: {year}-{month}
In large complex plant genomes, RNA-directed DNA methylation (RdDM) ensures that epigenetic silencing is maintained at the boundary between genes and flanking transposable elements. In maize, RdDM is dependent on Mediator of Paramutation1 (Mop1), a gene encoding a putative RNA dependent RNA polymerase. Here we show that although RdDM is essential for the maintenance of DNA methylation of a silenced MuDR transposon in maize, a loss of that methylation does not result in a restoration of activity. Instead, heritable maintenance of silencing is maintained by histone modifications. At one terminal inverted repeat (TIR) of this element, heritable silencing is mediated via histone H3 lysine 9 dimethylation (H3K9me2), and histone H3 lysine 27 dimethylation (H3K27me2), even in the absence of DNA methylation. At the second TIR, heritable silencing is mediated by histone H3 lysine 27 trimethylation (H3K27me3), a mark normally associated with somatically inherited gene silencing. We find that a brief exposure of high temperature in a mop1 mutant rapidly reverses both of these modifications in conjunction with a loss of transcriptional silencing. These reversals are heritable, even in mop1 wild-type progeny in which methylation is restored at both TIRs. These observations suggest that DNA methylation is neither necessary to maintain silencing, nor is it sufficient to initiate silencing once has been reversed. However, given that heritable reactivation only occurs in a mop1 mutant background, these observations suggest that DNA methylation is required to buffer the effects of environmental stress on transposable elements.
Most plant genomes are mostly transposable elements (TEs), most of which are held in check by modifications of both DNA and histones. The bulk of silenced TEs are associated with methylated DNA and histone H3 lysine 9 dimethylation (H3K9me2). In contrast, epigenetically silenced genes are often associated with histone lysine 27 trimethylation (H3K27me3). Although stress can affect each of these modifications, plants are generally competent to rapidly reset them following that stress. Here we demonstrate that although DNA methylation is not required to maintain silencing of the MuDR element, it is essential for preventing heat-induced, stable and heritable changes in both H3K9me2 and H3K27me3 at this element, and for concomitant changes in transcriptional activity. These finding suggest that RdDM acts to buffer the effects of heat on silenced transposable elements, and that a loss of DNA methylation under conditions of stress can have profound and long-lasting effects on epigenetic silencing in maize.
Introduction
Transposable elements (TEs) are a ubiquitous feature of all genomes. They survive in large measure because they can out-replicate the rest of the genome [1]. As a consequence of that replication TEs can threaten the integrity of the host genome. In response to this threat, all forms of life have evolved mechanisms by which TEs can be silenced when they are recognized as such and, importantly, maintained in a silenced state over long periods of time, even when the initial trigger for silencing is no longer present [2–4]. Because plant genomes are largely composed of TEs, the majority of plant DNA is maintained in an epigenetically silent state [5]. Because they are the primary target of epigenetic silencing in plants, TEs are an excellent model for understanding the means by which particular DNA sequences are targeted for silencing, and for understanding the means by which silencing can be maintained from one generation to the next [6]. Finally, because TEs have proved to be exquisitely sensitive to a variety of stresses [7–9], they can also teach us a great deal about the relationship between stress and epigenetically encoded memory of stress response.
In plants, heritable epigenetic silencing of TEs is almost invariably associated with DNA methylation [10–12]. The vast bulk of TEs in plant genomes are methylated and, with some notable exceptions [13], epigenetically silenced [14,15]. DNA methylation has a number of features that makes it an appealing mechanism by which silencing can be heritably propagated, either following cell divisions during somatic development, or transgenerationally, from one generation to the next. Because methylation in both the CG and CHG sequence contexts (where H = A, T or G) are symmetrical, information concerning prior DNA methylation can be easily propagated by methylating newly synthesized DNA strands using the parent strand as a template. For CG methylation, this is achieved by reading the methylated cytosine using VARIANT IN METHYLATION 1–3 (VIM1-3) [16,17] and writing new DNA methylation using the methyl transferase MET1 [18–20]. For CHG, methylation is read indirectly by recognition of H3K9 dimethylation (H3K9me2) by CMT3, which catalyzes methylation of newly synthesized DNA, which in turn triggers methylation of H3K9 [21–23].
Maintenance methylation of most CHH involves RNA-directed DNA methylation (RdDM). The primary signal for de novo methylation of newly synthesized DNA from previously methylated DNA sequences is thought to be transcription by RNA POLYMERASE IV (POLIV) of short transcripts from previously methylated templates [24–26]. This results in the production of small RNAs that are tethered to the target DNA by RNA POLYMERASE V (POLV), which is targeted by SU(VAR)3-9 homologs SUVH2 and SUVH9, which bind to methylated DNA [27]. This in turn triggers de novo methylation of newly synthesized DNA strands using the methyl transferases DRMT1/2 [28,29]. In addition to the RdDM pathway, CHH methylation can also be maintained due to the activity of CHROMOMETHYLASE2 (CMT2), which, similar to CMT3, works in conjunction with H3K9me2 to methylate non-CG cytosines, particularly in deeply heterochromatic regions of the genome [30]. Finally, because both histones and DNA must be accessible in order to be modified, chromatin remodelers such as DDM1 are also often required for successful maintenance of TE silencing [23,31]. In plants, effective silencing of TEs requires coordination between DNA methylation and histone modifications [32]. Together, these pathways can in large part explain heritable propagation of both DNA methylation and histone modification of TEs.
In large genomes such as that of maize, much of RdDM activity is focused not on deeply silenced heterochromatin, which is often concentrated in pericentromeric regions, but on regions immediately adjacent to genes, referred to as “CHH islands” because genes in maize are often immediately adjacent to silenced TEs [15,33]. In maize, mutations in components of the RdDM pathway affect both paramutation and transposon silencing [34]. Mutations in Mediator of Paramutation 1 (Mop1), a homolog of RNA DEPENDENT RNA POLYMERASE2 (RDR2), result in the loss of nearly all 24 nucleotide small RNAs, as well as the CHH methylation that is associated with them [35–37]. Despite this, mop1 has only minimal effects on gene expression in any tissue except the meristem [33,38], and the plants are largely phenotypically normal [39]. This, along with similar observations in Arabidopsis, has led to the suggestion that the primary role of RdDM is to reinforce boundaries between genes and adjacent TEs, rather than to regulate gene expression [33]. However, it should be noted that the mop1 mutation can in some cases have effects on plant phenotype [40]. Further, mop1 mutants can enhance the effects of exogenously applied ABA [41] and mutants of Required to maintain repression6 (Rmr6), a homolog of the PolIV subunit DNA-directed RNA polymerase IV subunit 1 (NRPD1) [42], are altered in their response to drought, suggesting that the RdDM pathway may play a role in buffering stress responses in maize [43,44]. Further, even in wild-type backgrounds, there is evidence that the process of heritable paramutation of an allele of R1, which is known to be dependent on RdDM, is sensitive to changes in temperature and light during specific stages of development [45].
Unlike animals, plants do not experience a global wave of DNA demethylation either in the germinal cells of the gametophyte or in the early embryo [46]. Thus, DNA methylation and associated histone modifications are an attractive mechanism for transgenerationally propagated silencing. Indeed, there is strong evidence that mutants that trigger a global loss of methylation can cause heritable reactivation of previously silenced TEs, although it is worth noting that even in mutants in which the vast majority of DNA methylation has been lost, only a subset of TEs are transcriptionally reactivated [47,48], and DNA methylation of many TEs can be rapidly reestablished at many loci via RdDM in wild-type progenies of mutant plants, suggesting that memory propagated via DNA methylation can be restored due to the presence of small RNAs that can trigger de novo methylation of previously methylated sequences [49,50].
In contrast to TEs, most genes that are silenced during somatic development in plants are associated with H3K27 trimethylation (H3K27me3), which requires the activity of the polycomb complexes PRC2 and PRC1, which together catalyze H3K27 methylation and facilitate its heritable propagation [51–53]. In plants, H3K27me3 enrichment is generally associated with genes rather than TEs [54,55], and numerous developmental pathways require the proper deposition and maintenance of this modification [56,57]. The most well explored example of this involves epigenetic setting of FLOWERING LOCUS C (FLC), a negative regulator of flowering in Arabidopsis [58,59]. In a process known as vernalization, prolonged exposure to cold results in somatically heritable silencing of this gene, which in turn results in flowering under favorable conditions in the spring. Somatically heritable silencing of FLC is initially triggered by non-coding RNAs, which are involved in recruitment of components of PRC2, which catalyze H3K27me3, which in turn mediates a somatically heritable silent state [58]. Importantly, H3K27me3 at genes like FLC is erased each generation, both in pollen and in the early embryo [60–62]. The fact that H3K27me3 must be actively reset suggests that in the absence of this resetting, H3K27me3 in plants is competent to mediate transgenerational silencing but is normally prevented from doing so.
Dramatic differences in TE content between even closely related plant species suggest that despite the relative stability of TE silencing under laboratory conditions, TEs frequently escape silencing and proliferate in natural settings [63]. Stress, both biotic and abiotic can often trigger TE transcription and, at least in some cases, transposition [7,64–67]. Further, there is evidence that the association of TEs and genes can result in de novo stress induction of adjacent genes [64,68,69].
Because of its dramatic and global effects on both gene expression and protein stability, heat stress has attracted considerable attention, particularly with respect to heritable transmission of TE activity. Although heat stress can trigger somatically heritable changes in gene expression, there appear to be a variety of mechanisms to prevent or gradually ameliorate transgenerational transmission of those changes [70,71]. Thus, for instance, although the ONSEN retrotransposon is sensitive to heat, it is only in mutants in the RdDM pathway that transposed elements are transmitted to the next generation [9,72]. Given that various components of regulatory pathways that have evolved to regulate TEs are up-regulated in germinal lineages, it is not surprising that a defect in one of these pathways would lead to an enhancement in the number of germinally transmitted new insertions [73,74]. The observation that it is the combination of both heat and components of the RdDM pathway results in reactivation of TEs, rather than each by itself has led to the suggestion that a key role of RdDM is to prevent TE activation specifically under conditions of stress [9,75].
Similar experiments using silenced transgenes have demonstrated that double mutants of mom1 and ddm1 cause silenced transgenes as well as several TEs to be highly responsive to heat stress, and the observed reversal of silencing can be passed on to a subsequent generation, but only in mutant progeny [76]. It is also worth noting that in many cases of TE reactivation, silencing is rapidly re-established in wild-type progeny [77,78]. The degree to which this is the case likely depends on a variety of factors, from the copy number of a given element, its position within the genome, its mode of transposition and the presence or absence of trans-acting small RNAs targeting that TE [79].
Our model for epigenetic silencing is the Mutator system of transposons in maize. The Mutator system is a family of related elements that share similar, 200 bp terminal inverted repeats (TIRs) but that contain distinct internal sequences. Nonautonomous Mu elements can only transpose in the presence of the autonomous element, MuDR. MuDR is a member of the MULE superfamily of Class II cut and paste transposons [80,81]. In addition to being required for transposition, the 200 bp TIRs within MuDR elements serve as promoters for the two genes encoded by MuDR, mudrA, which encodes a transposase, and mudrB, which encodes a novel protein that is required for Mu element integration. Both genes are expressed at high levels in rapidly dividing cells, and expression of both of them is required for full activity of the Mutator system [82,83]. MURA, the protein produced by mudrA, is sufficient for somatic excision of Mu elements, which results in characteristically small revertant sectors in somatic tissue. MuDR elements can be heritably silenced when they are in the presence of Mu killer (Muk), a rearranged variant of MuDR whose transcript forms a hairpin that is processed into 21–22 nt small RNAs that directly trigger transcriptional gene silencing (TGS) of mudrA and indirectly trigger silencing of mudrB when it is in trans to mudrA [4,84]. Because Muk can be used to heritably silence MuDR through a simple cross, and because silencing of MuDR can be stably maintained after Muk is segregated away, the MuDR/Muk system is an excellent model for understanding both initiation and maintenance of silencing. Prior to exposure to Muk, MuDR is fully active and is not prone to spontaneous silencing [85]. After exposure, MuDR silencing is exceptionally stable over multiple generations [84].
When mudrA is silenced, DNA methylation in all three sequence contexts accumulates within the 5’ end of the TIR immediately adjacent to mudrA (TIRA) [86]. Methylation at the 5’ and 3’ portions of this TIR have distinctive causes and consequences. The 5’ end of the TIR is readily methylated in the absence of the transposase, but this methylation does not induce transcriptional silencing of mudrA [87]. Methylation in this end of TIRA is readily eliminated in the presence of functional transposase. However, the loss of methylation in a silenced element in this part of the TIRA does not result in heritable reactivation of a silenced element. In contrast, CG and CHG methylation in the 3’ portion of TIRA, which corresponds to the mudrA transcript as well as to Muk-derived 22 nt small RNAs that trigger silencing, is not eliminated in the presence of active transposase and is specifically associated with heritable transcriptional silencing of mudrA.
The second gene encoded by MuDR elements, mudrB, is also silenced by Muk, but the trajectory of silencing of this gene is entirely distinct, despite the fact that the Muk hairpin has near sequence identity to the TIR adjacent to mudrB (TIRB) [4,84]. By the immature ear stage of growth in F 1 plants that carry both MuDR and Muk, mudrA is transcriptionally silenced and densely methylated. In contrast, mudrB in intact elements remains transcriptionally active in this tissue, but its transcript is not polyadenylated. It is only in the next generation that steady state levels of transcript become undetectable. Further, experiments using deletion derivatives of MuDR that carry only mudrB are not silenced by Muk when they are on their own, or when they are in trans to an intact MuDR element that is being silenced by Muk. This suggests that heritable silencing of mudrB is triggered by the small RNAs that target mudrA, but the means by which this occurs is indirect and involves spreading of silencing information from mudrA to mudrB.
Silencing of mudrA can be destabilized by the mop1 mutant. MOP1 is homolog of RDR2 that is required for the production of the vast bulk of 24 nt small RNAs in maize, including those targeting Mu TIRs [35–37,88]. However, silencing of MuDR by Muk is unimpeded in a mop1 mutant background, likely because Muk-derived small RNAs are not dependent on mop1 [89]. Further, although reversal of silencing of MuDR in a mop1 mutant background does occur, it only occurs gradually, over multiple generations, and only affects mudrA. In contrast, mudrB is not reactivated in this mutant background and, because mudrB is required for insertional activity, although these reactivated elements can excise during somatic development, they cannot insert into new positions.
[END]
[1] Url:
https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1009326
(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL:
https://creativecommons.org/licenses/by/4.0/
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
