Transgenerational Transmission of Epigenetic Marks During Reproduction in Arabidopsis

Abstract

Epigenetics has described non-DNA sequence-based transgenerational inheritance, in which phenotypic traits acquired over a lifetime by parents are passed to their progeny across several generations. Transgenerational epigenetic inheritance may be achieved by transmitting repressive epigenetic marks such as DNA methylation and histone 3 lysine 27 trimethylation (H3K27me3) in parental chromatin through sexual reproduction. In general, with infrequent exceptions, epigenetically modified architectures in parental chromatin are extensively erased by reprogramming during reproduction, leaving little possibility of being inherited by offspring. In comparison, plants exhibit transgenerational epigenetic inheritance with relatively greater frequency, although the underlying molecular mechanisms have remained unclear. Recent studies with the flowering plant Arabidopsis (Arabidopsis thaliana) have identified plant-specific mechanisms enabling the transgenerational transmission of epigenetic marks during reproduction. Arabidopsis achieves de novo chromatin methylation through the small interfering RNA (siRNA)-directed DNA methylation (RdDM) pathways. In addition, Polycomb repressive complex 2 (PRC2) catalyzes H3K27me3 deposition on exchangeable histone variants during reproduction in a plant-specific manner. This review describes recent progress in Arabidopsis research of transgenerational epigenetic inheritance, focusing on transmitting epigenetic marks through reproduction steps: meiosis, gametogenesis, and embryogenesis.

Introduction

Phenotypic information acquired throughout a parent’s life can be transferred to their progeny across several generations. Such non-DNA sequence-based transgenerational inheritance has been observed among many taxonomic groups, from yeast and invertebrates to plants and mammals, including humans [1,2]. Recently, epigenetics has described the inheritance mechanisms with transmission of the repressive epigenetic marks such as DNA methylation and histone 3 lysine 27 trimethylation (H3K27me3) in parental chromatin through sexual reproduction [3].

Plants exhibit transgenerational epigenetic inheritance much more frequently than mammals, as reported across multiple plant species in response to adverse environmental challenges such as drought, salinity, extreme temperatures, and pathogen infections [4]. Plants are sessile and therefore confront adverse environmental conditions by regulating the expression of resistance genes and activator/repressor genes by altering their chromatin architecture to expedite tolerance responses. Such chromatin remodeling involves complex interactions among several epigenetic marks, including DNA methylation, histone modifications, and non-coding RNAs [5]. Changes in the chromatin architecture of parents can be propagated to the next generation via sexual reproduction. However, mammals exhibit a high fidelity in epigenetic mark reprogramming, effectively resetting stress-induced changes in chromatin architecture during reproduction. In contrast, flowering plants transmit such epigenetic changes more easily, transferring acquired stress resistance to their progeny across multiple generations [6].

Arabidopsis (Arabidopsis thaliana) is an ideal model flowering plant for studying transgenerational inheritance due to its short life cycle (8–10 weeks) and small genome size (~135 Mb). Arabidopsis reproduction is initiated by generating pollen mother cells and a megaspore mother cell from adult somatic cells that differentiate from the reproductive shoot apical meristem (Figure 1) [7,8]. The meiotic competent cells (meiocytes) in these mother cells undergo meiosis to produce microspores and a megaspore, respectively. The microspores proceed to mitotic division to generate a mature male gametophyte, constituting two sperm cells within a companion vegetative cell. In parallel, the megaspore develops into the mature female gametophyte, which contains one egg cell within a central cell. The two sperm cells are double-fertilized by the egg and central cells, producing a zygote and an endosperm, respectively. The zygote develops into an embryo by embryogenesis to complete mature seed formation.

Figure 1.

Key steps in the sexual reproduction of Arabidopsis. From somatic cells in the male and female reproductive tissues, four diploid (2n) male meiocytes (pollen mother cells) and a female meiocyte (megaspore mother cell) are generated. Each male meiocyte produces haploid (n) microspores, and each microspore undergoes asymmetrical mitotic division to form a generative cell and a pollen vegetative cell (this step is omitted in this figure). The generative cell enters another round of haploid mitosis to generate two sperm cells within the vegetative cell, constituting the mature male gametophyte (pollen). The female meiocyte gives rise to a haploid (n) megaspore via meiosis, and the megaspore undergoes three rounds of mitotic cell division to generate an egg cell (n) and accessory cells within a central cell (2n), constituting the mature female gametophyte. The egg and central cells are double-fertilized by a sperm cell to produce a zygote (2n) and a triploid (3n) endosperm, respectively. The zygote develops into an embryo via embryogenesis to complete seed maturation. For more detailed information, refer to the references [7,8].

As described in this mini-review, recent studies with Arabidopsis have identified plant-specific mechanisms enabling the transgenerational transmission of epigenetic marks through meiosis, gametogenesis, and embryogenesis. Most eminently, small interfering RNAs (siRNAs) emerged as a cell-to-cell and long-distance mobile messenger of the germline-based transmission of DNA methylation status. Using plant-unique RNA polymerases, siRNAs mediate the RNA-directed DNA methylation (RdDM) pathways to achieve de novo DNA methylation during reproduction [9]. In parallel, the repressive epigenetic mark H3K27me3 has attracted attention as a potential mediator of the transmission of chromatin-based epigenetic information [10]. Deposition of H3K27me3 marks is catalyzed by Polycomb repressive complex 2 (PRC2) on exchangeable histone variants during reproduction in a plant-specific manner.

Transmission of DNA Methylation

Genomic DNA Methylation

High-resolution DNA methylation mapping of the entire Arabidopsis genome has revealed that cytosine bases are often methylated mainly in the form of 5-methylcytosine (5-mC) at the CG, CHG, and CHH sites (where H represents A, T, or C) [11]. CG is the predominant site for methylation, which is catalyzed by DNA METHYLTRANSFERASE 1 (MET1) [12]. The Arabidopsis mutant met1-3 formed progressively aberrant epigenetic patterns over several generations [13]. CHH cytosines are methylated by the de novo DNA methyltransferases DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) and CHROMOMETHYLASE 2 (CMT2). These two enzymes can also methylate cytosines in CHG sites. Inversely, DNA demethylation is catalyzed by 5-mC DNA glycosylases such as DEMETER (DME) and REPRESSOR OF SILENCING 1 (ROS1) [14]. In Arabidopsis heterochromatin, nearly all endogenous genes and TEs are found in a transcriptionally silenced state maintained by DNA methylation [15].

In Arabidopsis, de novo DNA methylation is achieved through the RdDM pathways [16,17]. Canonical RdDM relies on transcriptional mechanisms led by two plant-specific RNA polymerases (Pols), Pol IV and Pol V (Figure 2) [18]. In brief, Pol IV produces a single-stranded RNA transcript from an endogenous gene or a TE. Then, RNA-dependent RNA polymerase RDR2 transcribes a second strand using the RNA transcript as a template. The double-stranded RNA duplex is cleaved by the Dicer-like ribonuclease DCL3 to produce 24-nt siRNAs. A single siRNA is loaded onto an Argonaute family protein AGO4. Then, Pol V transcribes a nascent long non-coding RNA (lncRNA) transcript in a specific genomic region that base pairs with the AGO4-bound siRNA. DRM2 catalyzes de novo cytosine methylation of the corresponding DNA template from which the siRNA was derived.

Figure 2.

RNA-dependent DNA methylation (RdDM) pathways in Arabidopsis. In the canonical RdDM pathway, RNA polymerase IV (Pol IV) is recruited through interactions with SAWADEE HOMEODOMAIN HOMOLOGUE1 (SHH1) to the target region and transcribes a single-stranded RNA (ssRNA) from genes or transposable elements (TEs). The ssRNA is copied into a double-stranded RNA (dsRNA) by RNA-DEPENDENT RNA POLYMERASE2 (RDR2) with the assistance of the chromatin remodeler CLASSY1 (CLSY1). The dsRNA is processed by DICER-LIKE3 (DCL3) into 24-nucleotide siRNAs and incorporated into ARGONAUTE4 (AGO4). AGO4 is recruited through interaction with the C-terminal domain of Pol V, and then Pol V transcribes a nascent long non-coding RNA (lncRNA) transcript that base-pairs with AGO4-bound siRNAs. Then, RNA-DIRECTED DNA METHYLATION 1 (RDM1) links AGO4 and DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2), which catalyzes de novo cytosine methylation of the corresponding DNA and thereby mediates transcriptional gene silencing (TGS). In noncanonical pathways, Pol II, RDR6, and DCL2/4 mediate the generation of 21- and 22-nt siRNAs. These siRNAs interact with AGO4/6 and enter the Pol V-dependent process as described in the canonical pathway. In addition, some of the 21–22-nt siRNAs are loaded onto AGO1/7 and mediate post-transcriptional silencing in the cytoplasm. DNAs are black and RNAs are blue. The figure omits other factors associated with each step to emphasize only the main steps of RdDM pathways. For more detailed information, refer to the references [16,17].

In addition to the canonical RdDM that triggers promoter methylation and transcriptional gene silencing (TGS), several alternative noncanonical pathways linked to post-transcriptionally silenced (PTGS) genes have been identified [19,20]. For instance, at some TAS loci encoding trans-acting siRNAs or newly integrated TEs, the normal polymerase Pol II transcribes noncoding RNAs (ncRNAs). From these ncRNAs, RDR6 and DCL2/4 generate 21- and 22-nt siRNAs [21,22]. These siRNAs are incorporated into AGO4/6 and enter the Pol V-dependent process described in the canonical pathway to reinforce TGS. In addition, some of the 21-22-nt siRNAs are loaded onto AGO1/7 and mediate PTGS through the methylation within protein-coding regions and cleavage of complementary target mRNAs in the cytoplasm.

Transgenerational Inheritance of DNA Methylation

Genomic DNA methylation is the clearest line of evidence demonstrating transgenerational epigenetic inheritance. Cortijo et al. [23] analyzed a population of isogenic Arabidopsis lines that segregate experimentally induced DNA methylation changes at hundreds of genome regions and identified several differentially methylated regions (DMRs). These DMRs act as epigenetic quantitative trait loci (QTLs) that explain 60–90% of the heritability related to flowering time and primary root length. Johannes et al. [24] generated epigenetic recombinant inbred lines (epiRILs) in Arabidopsis by crossing two parents with contrasting DNA methylation profiles. These epiRILs exhibited stable heritability for flowering time and plant height over at least eight generations.

siRNAs of diverse sizes (21–24-nt) derived from gene or TE transcripts can move by cell-to-cell trafficking via plasmodesmata and long-distance travel through phloem to direct methylation to identical DNA sequences [25]. Arabidopsis mutants with impaired siRNA biogenesis (dicer-like2/dicer-like3/dicer-like4) failed to exhibit inherited resistance to caterpillar herbivory [26]. Remarkably, Long et al. [27] illustrated that 24-nt siRNAs transcribed from hypermethylated TEs in meiocyte nurse cells are transported into meiocytes to reconstitute germline DNA methylation patterns in Arabidopsis.

Transmission of DNA Methylation During Reproduction

In Arabidopsis, an overall reduction of DNA methylation occurs in the male germline during gametogenesis, partly due to demethylase activities exerted by DME and ROS1 [28,29]. CG and CHG methylations are retained at high levels in microspores and sperm cells, but CHH methylation is significantly reduced. The CHH methylation levels increase progressively but remain lower in sperm than in somatic cells. In vegetative cells, CG methylation becomes lower than in sperm, but CHH methylation in heterochromatic TEs is restored by RdDM [30].

Slotkin et al. [31] found that TEs are reactivated and transposed in the Arabidopsis vegetative nucleus. The vegetative cell undergoes a programmed decondensation of heterochromatin [32]. Then, chromatin remodeler CLASSY (CLSY) family proteins recruit Pol IV to distinct genomic targets, facilitating locus-specific regulation of the RdDM pathway [33]. Thereby, the vegetative nucleus reactivates TE transcription and produces siRNAs from the TE transcripts. The mobile siRNAs are transferred to sperm cells and possibly loaded onto AGOs preferentially enriched in these male gametes [34]. As a result, DNA methylation is reinforced by the RdDM and suppresses the mutagenic activity of TEs, which provides genome stability in the sperm cells.

Ingouff et al. [35] developed sensors reporting with single-cell resolution and observed in a fluorescent reporter assay that CG and CHH methylation levels remain stable throughout female gametogenesis. CHH methylation in the egg cell depends on DRM2 and Pol V, not Pol IV. Thus, DNA methylation within the egg cell appears to rely on a Pol II-initiated noncanonical RdDM pathway or reception of exogenous siRNAs without production. Supporting the speculation, Erdmann et al. [36] observed in a fluorescent reporter injection assay that siRNAs move from the central cell to the egg cell to reinforce TE silencing. After fertilization, the zygote genome is still partially demethylated in the early stages; however, during embryogenesis, intensive RdDM restores methylation levels in adult somatic tissues [37].

Transmission of Histone Marks

H3K27me3 Mark Deposition

Along with DNA methylation, H3K27me3 is an important chromatin-based epigenetic mark for gene silencing. In Drosophila, the level of H3K27 methylation is regulated by PcG protein complexes. One member of the PcG protein complexes, Polycomb repressive complex 2 (PRC2), contains an enzymatic subunit, ENHANCER OF ZESTE HOMOLOG 1 (Ezh1) or Ezh2, that exerts histone methyltransferase activity on H3K27 [38]. H3K27me3 deposited in target loci recruits another PcG protein complex, PRC1, which compacts chromatin nucleosomes to provide a major regulatory mechanism for suppressing unnecessary gene expression or TE activation. The repressive function of PRC2 is counteracted by the H3K27me3 demethylase activity of Jumonji (JMJ) C-domain proteins [39].

In Drosophila, PRC2 recruitment to target genes depends on the coordinated action of DNA-binding proteins and transcription factors that recognize Polycomb group response elements (PREs) [40]. PREs are cis-regulatory DNA elements of several hundred base pairs in length located in the promoter regions. Homologs of Drosophila PcG components, interactors, and their target genes have been identified in the Arabidopsis genome [41]. In addition, several JMJ-type histone demethylases, such as EARLY FLOWERING 6 (ELF6) and RELATIVE OF EARLY FLOWERING 6 (REF6), were identified in Arabidopsis [42].

Transgenerational Inheritance of H3K27me3 Marks

In mammals, chromatins in the paternal genome are extensively reprogrammed to exchange histones with protamines during reproduction [43]. No histone-protamine exchange occurs in the plant genome. Instead, histones are replaced by newly synthesized histone variants. Ingouff et al. [44] performed transcriptional profiling and live imaging of H3 variants encoded by the Arabidopsis genome during reproduction. H3.1 variants were detectable in the male and female gamete precursors but disappeared rapidly before sperm and egg cell mitosis. Instead, H3.3 variants represented the main class of H3 proteins present in male and female gametes. However, upon fertilization, the H3 variants derived from male and female gametes were replaced with de novo synthesized H3 variants in zygote chromatin.

The canonical variant H3.1 is synthesized during S-phase and deposited into the chromatin through a DNA replication-coupled pathway. Thus, the disappearance of H3.1 variants before sperm and egg cell mitosis is merely due to arrested cell-cycle progression. Trimethylation at K27 on H3.1 variants maintains Polycomb-mediated gene silencing in plants, which is essential for transmitting epigenetic information during mitotic cell division [45]. In a study with human cells, chromatin immunoprecipitation of epitope-tagged H3.3 and array hybridization revealed that H3.3 variants are coenriched with H3K27me3, and incorporate into nucleosomal chromatin in a replication-independent manner, both during and outside S-phase [46]. Therefore, replacement of H3.3 carrying H3K27me3 with newly synthesized histones in zygote chromatin has been interpreted as a mechanism for blocking the transgenerational transmission of H3K27me3-dependent epigenetic information.

Nevertheless, transgenerational inheritance of H3K27me3 marks has been observed in the mutation studies of Jumonji-type histone demethylases in Arabidopsis. Crevillén et al. [47] observed that a mutation in ELF6 impaired reactivation of the floral repressor locus FLOWERING LOCUS C (FLC) in reproductive tissues, resulting in transgenerational inheritance of a partially vernalized state. In the next generation, higher levels of H3K27me3 at the FLC locus and lower FLC expression were observed in this mutant compared to wild-type plants. Antunez-Sanchez et al. [48] generated the epimutant lines elf6-C/ref6-5 by reciprocal cross with wild-type Arabidopsis plants. They observed that pleiotropic developmental defects exhibited in the epimutant were stably inherited over multiple generations. It was proposed that some H3K27me3 marks may not be reset if they are distal to the target sequences recognized by the demethylases and could be inherited across generations. In addition, this study observed that failure in H3K27me3 resetting caused the loss of DNA methylation at heterochromatic loci, leading to TE activation. The significance underlying the dual roles of histone demethylases in maintaining genome integrity and transcriptional states remains to be further investigated.

Transmission of H3K27me3 Marks During Reproduction

Borg et al. [49] performed inflorescence mark line analysis with Arabidopsis and detected H3K27me3 in the microspore, implicating transmission from the microspore mother cell through meiosis. Sperm chromatin incorporates the sperm-specific histone variant H3.10, which cannot be methylated at K27, in addition to H3.3. Additionally, PRC2 components, including histone methyltransferase-encoding subunits, and most PRC1 subunits were undetectable during sperm development. In a JMJ demethylase mutant elf6/ref6/jmj13, H3K27me3 levels were increased in the sperm cells, indicating that active demethylation also contributes to paternal H3K27me3 resetting. Thus, H3K27me3 is globally lost from histone-based sperm chromatin in Arabidopsis. In contrast, expression of the Polycomb machinery was unaltered, and no major loss of H3K27me3 was observed in female gametophytes.

Zenk et al. [50] observed that maternally inherited H3K27me3 is propagated in the early Drosophila embryo and regulates the activation of enhancers and lineage-specific genes during development. Similarly, in plants, the H3K27me3 marks appear to be transmitted through the maternal-to-zygotic transition, inherited by the embryo, and maintained throughout the nuclear division of somatic cells in the next generation.

Conclusions and Outlook

As described in this review, parental epigenetic information can be preserved and reconstituted during reproduction, enabling the transgenerational inheritance of acquired traits. siRNA-mediated RdDM and PRC2-dependent H3K27me3 deposition have been identified as the main mechanisms of epigenetic inheritance in Arabidopsis. However, the molecular mechanism underlying epigenetic inheritance remains an important research challenge. Importantly, the mechanism by which epigenetic marks in chromatin are disassembled and reassembled during the passage of mitotic cell division and meiotic transmission remains speculative. Additionally, it is entirely unknown how complex, interrelated epigenetic transformation networks accumulated in tissues and organs of different cell types can be collected and transmitted to a few reproductive cells. In this regard, Alexandran [51] recently proposed the existence of pluripotent adult stem cells with a unique ability to dedifferentiate/differentiate, or transdifferentiate to different cell types, although this hypothesis awaits experimental validation. Extending our perspective, more fundamental propositions are needed to address the inheritance of acquired traits. For instance, whether transgenerational epigenetic inheritance can contribute to evolution, ie, support Lamarck’s transmutation theory, remains controversial. Advances in innovative experimental tools are expected to improve our understanding and practical applications of the molecular mechanisms underlying transgenerational epigenetic inheritance. For example, the MethylC-sequencing library preparation method enables the genome-wide identification of DNA methylation states at single-base resolution [52]. Chromatin immunoprecipitation followed by sequencing (ChIP-seq) has facilitated genome-wide profiling of histone modifications. Notably, an efficient method for histone modification profiling at single-cell resolution was recently developed, enabling different cell types to be distinguished [53].

Glossary

H3K27me3

Histone 3 lysine 27 trimethylation

DCL

Dicer-like ribonuclease

PcG

Polycomb Group

PRC

Polycomb Repressive Complex

RdDM

RNA-directed DNA Methylation

siRNA

Small Interfering RNA

TE

Transposable Element

Author Contributions

JJC wrote and edited the manuscript.

Funding Statement

This work was supported by the National Research Foundation of Korea (Grant number NRF-2020R1I1A1A01070089).