Transposons are highly conserved in plants and have created a symbiotic relationship with the host genome. An important factor of the successful communication between transposons and host plants is epigenetic modifications including DNA methylation and the modifications of the histone tail. In plants, small interfering RNAs (siRNAs) are responsible for RNA-directed DNA methylation (RdDM) that suppresses transposon activities. Although most transposons are silent in their host plants, certain genomic shocks, such as an environmental stress or a hybridization event, might trigger transposon activation. Further, since transposons can affect the regulation mechanisms of host genes, it is possible that transposons have co-evolved as an important mechanism for plant development and adaptation. Recent new findings reveal that siRNAs control not only transcriptional activation, but also suppress transgenerational transposition of mobile elements making siRNAs critically important towards maintaining genome stability. Together these data suggest host-mediated siRNA regulation of transposons appears to have been adapted for controlling essential systems of plant development, morphogenesis, and reproduction.
Transposons of various classes, including both DNA transposons and retrotransposons, are abundant in plant genomes (Feschotte et al. 2002). The copy number of transposons influences the genome size. For example, a massive number of transposons make up more than 85% of the approximately 2300 MB maize genome (Wilson et al. 2009). The most abundant families of transposons in higher plants are long terminal repeat (LTR)-type retrotransposons (Kumar & Bennetzen 1999). In contrast to the “cut and paste” mechanism of DNA transposons use for transposition, retrotransposons transpose via an RNA intermediate that is reverse transcribed into extrachromosomal DNA and inserted into the genome. In addition to LTR-type families, Non-LTR retrotransposons were identified as ubiquitous components of nuclear genomes in many species. Non-LTR retrotransposons are divided into long interspersed nuclear elements (LINEs) and short interspersed nuclear elements (SINEs). LINEs are able to transpose autonomously while non-autonomous SINEs depend on the reverse transcription machinery of other retrotransposons. Although a large number of retrotransposons are conserved in intergenic regions (SanMiguel et al. 1996; Kumar & Bennetzen 1999) several families are conserved in gene rich regions (Kumar 1996; Kumar & Bennetzen 1999) possibly indicating an insertion preference of the retrotransposon or an evolutionary advantage at integration sites. This is surprising since active transposons are highly mutagenic and not only change gene expression, but also cause chromosome breakage and recombination leading to genomic instability. Some insertions, however, may favorably affect gene splicing or act as an enhancer or a promoter to nearby genes (Girard & Freeling 1999). Despite the potential to either harm or benefit the host genome, in nature most transposons are silent and rarely transpose due to the point mutations, deletions, or recombination that abolish their activities. In the event that full-length autonomous transposons are intact and can potentially transpose, host plants have evolved various types of epigenetic regulation to defend from transposition. DNA methylation is one regulatory mechanism that suppresses transcriptional activation, which is established by the RNA directed DNA methylation (RdDM) pathway guided by small RNAs (Lister et al. 2008). Indeed, about 37% of the methylated loci were associated with siRNA clusters (Zhang et al. 2006) in the Arabidopsis genome. Recently, siRNAs have also been revealed to play important roles for transposon inactivation, not only at the transcriptional level, but also at the transpositional level. The recent discoveries of plant-specific transposon regulation involving small RNAs in plants are reviewed in this article.
An siRNA biogenesis and RNA-directed de novo methylation in plants
RNA interference (RNAi) is a mechanism by which double strand RNA (dsRNA) is cleaved by members of the dicer-family proteins into short 21 to 30 nucleotide (nt) small RNAs (siRNAs) that can be channeled into different pathways. Higher plants evolved specific DNA-dependent RNA polymerases (Kanno et al. 2005; Onodera et al. 2005; Pontier et al. 2005; Mosher & Melnyk 2010) RNA polymerase IV and V (Pol IV and Pol V), which are plant-specific homologues of RNA polymerase II, to produce initial RNA transcripts for RNA silencing and siRNA-induced methylation, respectively (Matzke & Birchler 2005; Ream et al. 2009; Schnable et al. 2009). In Arabidopsis, single stranded Pol IV transcripts are transcribed by RNA-dependent RNA polymerase 2 (RDR2) to make double-stranded precursors that are processed by DICER-LIKE 3 (DCL3) to generate 24 to 26 nt siRNAs (Pontier et al. 2005; Zhang et al. 2007; Mosher et al. 2008). With an analogous function to the fission yeast RITS complex, the RNA-induced silencing complex (RISC) containing ARGONAUTE 4 (AGO4) binds a subset of these siRNAs (Qi et al. 2006). After being loaded with a siRNA, AGO4 interacts with Pol V to recruit the DNA methyltransferase, DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) (Cao & Jacobsen 2002; Matzke et al. 2004; Matzke & Birchler 2005), leading to de novo methylation the target.
Transposon-associated siRNAs and gene regulations in Arabidopsis
Transposon-derived siRNAs may introduce DNA methylation of nearby genes via RdDM and influence gene expression. One well characterized example is FLOWERING LOCUS C (FLC), a major repressor of flowering, that is negatively regulated by vernalization leading to natural variation in flowering behavior among Arabidopsis accessions (Boss et al. 2004). The mechanism for this phenotype is associated with a 1224 bp non-autonomous Mutator-like transposon insertion in the first intron of FLC (Gazzani et al. 2003; Michaels et al. 2003) acting in cis to reduce expression of the FLC allele in the accession Landsberg erecta (Ler) (Liu et al. 2004). The inserted transposon also makes FLC subject to repressive chromatin modifications mediated by siRNAs generated from unlinked, homologous transposable elements elsewhere in the genome.
Another example is the imprinted gene FLOWERING WAGENINGEN (FWA) that is specifically expressed in the endosperm but is silent in vegetative tissues of A. thaliana. The tissue-specific imprinted expression of FWA depends on DNA demethylation of the FWA promoter, which is comprised of two direct repeats containing a sequence related to a short interspersed nuclear (SINE) retrotransposon (Kakutani et al. 2007). If the FWA promoter is methylated, localized heterochromatin is established leading to transcriptional silencing and the biogenesis of small interfering RNAs from these SINE-related tandem repeats (Lippman et al. 2004; Chan et al. 2006). Similar to the inserted transposon of FLC described above, these siRNA can also target the RdDM activity to establish de novo silencing at unmethylated FWA transgenes (Cao & Jacobsen 2002; Chan et al. 2004).
Transposon silencing in plant gametes and embryogenesis
Plants and animals use a different system for fertilization and the sexual reproduction involved haploid gametes (Walbot & Evans 2003). In plants, spore mother cells (SMCs) that are destined to undergo meiosis are specified late during the development of the diploid generation of the life cycle, the sporophyte. In flowering plants this occurs in anthers and ovules, respectively. The meiotic products in plants, the spores, do not differentiate directly into gametes as in animals; rather, they divide mitotically to produce multicellular, haploid gametophytes. In flowering plants, the male gametophyte (pollen) consists of three cells that include two gametic sperm cells harbored within an accessory and vegetative cell responsible to deliver the sperm cells to the female gametes. The female gametophyte (embryo sac) produces two gametes, the egg and central cell, and five accessory cells in which two synergids assisting fertilization and three antipodals. Double fertilization generates two fertilization products with distinct developmental fates. While the fertilized egg gives rise to the embryo, the fertilized central cell generates an extra-embryonic nurturing tissue, the endosperm (Rudall 2006; Rudall & Bateman 2007; Dumas & Rogowsky 2008; Baroux et al. 2011).
Suppression of transposon activity in gametes is an important defense to ensure host genome stability by preventing active transposition being transmitted to the next generation. Biogenesis of Pol IV-dependent (p4)-siRNAs from thousands of loci, corresponding to more than 1% of the Arabidopsis genome, including repetitive elements and transposons (Zhang et al. 2007; Mosher et al. 2008) can accumulate in specific tissues, such as the developing endosperm (Mosher et al. 2009). Interestingly, the expression of p4-siRNAs in developing endosperm specifically originates from maternal chromosomes indicating genome imprinting (Mosher et al. 2009). The siRNAs produced in the endosperm have been reported to move into the egg cell to guide DNA methylation, seemingly to reinforce silencing in the germ cells (Feng et al. 2010). While this mechanism likely functions in suppressing transposon activity in the embryo, the derepression of transposons is not considered harmful since the endosperm is not inherited by the next generation.
Another important role of silencing in the control of female gamete formation was revealed by the action of ARGONAUTE 9 (AGO9) in Arabidopsis. AGO9 is highly expressed in ovules and anthers of Arabidopsis (Olmedo-Monfil et al. 2010). Interestingly, AGO9 expresses in cytoplasmic foci of somatic companion cells both outside and before differentiation of the gamete lineage (Olmedo-Monfil et al. 2010). In ago9 mutants, some transposons are activated indicating that AGO9 is necessary for transposon silencing in the ovule and an endogenous 24 nt siRNA biosynthetic pathway may play an important role for AGO9-dependent transposon suppression. In this model, epidermal cells in the young ovule primordium produce p4-siRNAs that are templates for AGO9 to generate siRNA signals. These signals may be channeled into a secondary siRNA pathway involving RDR6 and SGS3 to move the siRNA to sub-epidermal nucellar cells to inactivate transposons. Additionally, this function restricts the surrounding cells from ectopic development of a gamete lineage (Olmedo-Monfil et al. 2010). Thus, AGO9 controls gametic cell commitment by acting in a non-cell-autonomous small RNA-dependent pathway in the developing ovule where transposon regulation remains of central importance.
In Arabidopsis, epigenetic reprogramming also plays an important role for the transposon silencing in pollen (Slotkin et al. 2009). Many of the genes involved in siRNA biogenesis and silencing are either not expressed or expressed at low levels in pollen; however, the chromatin remodeling ATPase DECREASE IN DNA METHYLATION 1 (DDM1) (Brzeski & Jerzmanowski 2003) accumulates specifically in sperm cells, but not in the vegetative nucleus of mature pollen (Slotkin et al. 2009). In wild type vegetative nuclei, the reduction of DDM1 correlates with DNA demethylation and transposon reactivation. Much like the endosperm, transposon activation in pollen vegetative nuclei does not impair fitness of the next generation since the vegetative nucleus does not contribute DNA to the fertilized embryo or endosperm. It has been postulated that the benefit of deregulation within the vegetative nucleus occurs when small RNA are transported from the pollen grain cytoplasm to within the sperm cells resulting in siRNA accumulation to reinforce silencing of transposons in the gametes (Slotkin et al. 2009).
An siRNA and transposon mobility in Arabidopsis
In Arabidopsis, 24 nt small RNAs were most abundant, representing 58% of the unique small RNAs, followed by 23 nt (18%), 22 nt (10%), and 21 nt (9%) small RNAs (Ha et al. 2009). However, mutants compromised in 24 nt siRNA biogenesis exhibited no morphological defects despite transcriptional reactivation of some transposons (Herr et al. 2005; Kanno et al. 2005; Onodera et al. 2005; Pontier et al. 2005). This finding questioned the role of siRNAs in controlling transposon mobility. Recently, insight into the regulation of transposon mobility was reported for a Copia93 retrotransposon family member, named “Evade” (EVD). Although EVD transcription was suppressed by CG methylation, transpositional activity was independently regulated through a post-transcriptional mechanism involving siRNA and methylation of H3K9 (Mirouze et al. 2009). CG methylation in Arabidopsis is maintained by the MET1 methyltransferase (Law & Jacobsen 2010) and its depletion affects the distribution of methylation in genes and transposons (Ronemus et al. 1996; Kato et al. 2003). In met1 mutants, Evade transcriptional activity was released, yet EVD did not transpose. However, when met1 mutant was combined with a mutant deficient in siRNA biogenesis or the H3K9 histone methyltransferase KRYPTONITE, a burst of Evade transposition was observed (Mirouze et al. 2009). Interestingly, this transpositional burst was limited to only EVD’s mobilization (Mirouze et al. 2009). In contrast, analysis of the ddm1 mutant, which affects both DNA methylation and histone methylation levels (Hirochika et al. 2000; Miura et al. 2001; Gendrel et al. 2002; Lippman et al. 2003; Tsukahara et al. 2009) revealed a wider release of retro-element mobilization during inbreeding (Herr et al. 2005). This indicates that epigenetic regulation of retrotransposition is controlled by selective machinery and specific regulatory systems may be required for different transposon targets.
Another copia-type retrotransposon in Arabidopsis, named ONSEN, was transcriptionally activated with heat stress (Ito et al. 2011) leading to the synthesis of extrachromosomal DNA that can potentially transpose. The accumulation of ONSEN transcript and extrachromosomal DNA was much higher in the mutants impaired in the siRNA biogenesis (Ito et al. 2011). Importantly, no transposition was observed in vegetative tissues of either the wild type or the siRNA biogenesis mutant, however, a high frequency of new insertions was observed in the progeny of stressed plants deficient in siRNAs (Ito et al. 2011). Analysis of the insertion patterns revealed that the transgenerational transposition occurred during flower development, with evidence for mobilization before gametogenesis, demonstrating that siRNA-mediated transpositional suppression is important not only in reproductive cells, but also in the somatic tissue that will produce the gametes (Ito et al. 2011). Although the exact spatial and temporal regulation of ONSEN transposition in flower requires further analysis, it is likely that siRNA inhibition of transposition occurs in a developmental or tissue-specific manner (Fig. 1).
Interspecific and natural variation of transposon-mediated siRNAs in plants
The horizontal transfer of a retrotransposon into a new host species allows further investigation into the regulatory mechanisms controlling transposon amplification (Perez-Hormaeche et al. 2008). In one study, the LTR retrotransposon Tnt1 in tobacco (Nicotiana tabacum) was introduced into Arabidopsis. As a result, 24 nt siRNAs were produced that targeted the promoter in the LTR region to establish non-CG methylation and transcriptional silencing. The suppression of Tnt1 was dependent on copy number given that the stable reversion of silencing was obtained when the number of Tnt1 elements was reduced by genetic segregation to two copies (Perez-Hormaeche et al. 2008). This suggests that the maintenance of transposon silencing established by a copy number “threshold” can be released in some circumstances.
A natural situation in which transposon copy number variation may occur is when newly formed interspecific hybrids or resynthesized allopolyploids are produced, which can cause a “genomic shock”. Such a shock can lead to genome-wide alteration of gene expression, including transposon activation (Ha et al. 2009). The distribution of siRNA in two closely related species, Arabidopsis thaliana and Arabidopsis arenosa, a natural allotetraploid of Arabidopsis suecica, and resynthesized allotetraploid lines derived from A. thaliana and A. arenosa were analyzed (Ha et al. 2009). The results suggested that small RNAs produced during interspecific hybridization or polyploidization serve as a buffer against the genomic shock occurring in interspecific hybrids and allopolyploids. Among 6000 siRNA-generating transposons in A. thaliana, 5123 (≈85%) also produced siRNAs in one or more allotetraploids. A. thaliana siRNA populations underwent rapid changes in nascent F1 allotetraploids, but were stably maintained through the seventh generation (F7) indicating that stable inheritance of transposon-associated siRNA maintains chromatin and genome stability.
How such stable inheritance might affect genome evolution was also examined in Arabidopsis thaliana and Arabidopsis lyrata, two related species, yet with major transposon families having more copies in A. lyrata (Hollister et al. 2011). The difference indicates that most transposons have either been more active in A. lyrata or that selection against gene expression patterns modified by transposons is more stringent in A. thaliana (Hollister et al. 2011). A comparison of the 24 nt siRNA complement between the two species revealed that siRNA-targeted transposons were associated with reduced gene expression within both species, but also created gene expression differences between the orthologues (Hollister et al. 2011). In addition, A. lyrata transposons were targeted by a lower fraction of uniquely matching siRNAs, which are associated with more effective silencing of the expression. These results indicate that the efficacy of RdDM silencing and transposon proliferation clearly differ between the two species. What remains to be determined is whether this indicates genome evolution requires tolerance to modified gene expression patterns induced by proliferating transposons, or conversely, evolution requires an ability to purge new insertions from the genome, or some combination of these two functions.
A locus-specific example of natural variation comes from the analysis of highly abundant 24 nt siRNAs found in the ecotype Ler that could direct RdDM and heterochromatinization towards hobo, Activator, Tam3 (hAT) (Rubin et al. 2001) transposons adjacent to the promoter of FLC (Zhai et al. 2008). Despite the same hAT element in ecotype Columbia (Col) with almost the identical DNA sequence, the low amount of siRNAs detected did not affect FLC activities. A genome-wide comparison of Ler and Col small RNAs identified at least 68 loci matched by a significant level of the 24 nt siRNAs present specifically in Ler, but not Col, with nearly half of the loci being related to repeat or transposon sequences. The analysis suggested that the same region could be led to a different epigenetic status because of the difference in their corresponding small RNA abundance and between the two closely related Arabidopsis ecotypes, supporting the model that small RNA-directed epigenetic differences may exist among natural populations.
Environment stress and siRNA-mediated control of transposons
Environmental conditions influence plant growth and development by changing the gene expression, which can include transposon regulation (Zeller et al. 2009). One example was shown in Arabidopsis where a cluster of small RNAs (smRPi1LTR) originating from the LTR of Copia95 retrotransposon was induced by phosphate (Pi) starvation (Hsieh et al. 2009). The smRPi1LTR is likely generated from the cleavage of a single-stranded RNA precursor, rather than a double-stranded RNA precursor because the passenger strand of its duplex was identified from the hairpin stem with two mismatched base pairs. The different expression pattern between Columbia and Landsberg accessions indicated that the Pi-responsible smRPi1LTR is a newly evolved small RNA due to rapid rearrangement of LTR and may represent an intermediate small RNA species transitioning from siRNAs to microRNAs (miRNAs). The Pi-responsive small RNAs and their target genes are likely involved in the development or regulation of adaptive response to phosphate starvation (Hsieh et al. 2009). A second example in maize was demonstrated in response to cold stress, resulting in downregulation of DNA methyltransferase (ZmMET1) in root tissues causing hypomethylation in an Ac/Ds transposon region. The demethylation was specific to this transposon region and primarily occurred in roots indicating that the methylation level was decreased selectively by cold stress (Steward et al. 2000). The transcriptional activation from the transposons seemed to trigger locus-specific siRNA-mediated RdDM because the stress-induced methylation did not affect global methylation levels.
The transposon Tam3 in Antirrhinum majus undergoes low temperature–dependent transposition (LTDT) in which Tam3 is activated at low growth temperatures of 15°C, whereas the activity is strictly suppressed at high growth temperatures of 25°C (Carpenter et al. 1987). The methylation level of the Tam3 sequence at 15°C is markedly lower than that at 25°C and the state is reversibly altered demonstrating that the temperature-dependent change can occur during the lifetime of a single plant (Hashida et al. 2003, 2006). Decreases in methylation occurred only in the tissues that underwent cell division after the decrease in temperature and the methylation state was stable in old tissues, even at a low temperature (Hashida et al. 2006). The observation indicated that the Tam3 methylation level in LTDT is regulated by Tam3 activity and that at low temperatures when siRNA-mediated methylation might decrease, re-expression of silenced transposons may occur.
More direct evidence of siRNA-mediated regulation has come from the ONSEN transposition. ONSEN was activated in heat-stress and transposed to the next generation in a mutant of siRNA biogenesis. It is notable that ONSEN was not activated in the plants treated with DNA methylation inhibitor 5-azacytidine or within ddm1 mutant plants (Ito et al. 2011). Although the activation of ONSEN is likely independent of DNA methylation, siRNA is required for the immobilization. The experiments demonstrated that the heat-induced retrotransposition in the second generation of stressed pol IV mutant had an impact on the transcriptional regulation of endogenous loci harboring new ONSEN insertion. The insertions under heat stress may drive adaptation towards developing new expression variants with ONSEN acting as a heat responsive promoter element at new insertion sites. Also a gene in the Columbia accession harboring a natural insertion of ONSEN was heat activated. To determine the physiological relevance of the activation, heat responsiveness was analyzed on the gene in the Zürich accession where ONSEN is absent at the location. The result showed that heat-induced activation in the Columbia accession was much more pronounced than in the Zürich accession (Ito et al. 2011). Together, these results provide multiple examples of rapid host responses to a range of environmental stresses that can allow transposons to alter genome structure and/or gene regulation, thereby influencing future generations, which is of great importance to further study the connection between the stability of acquired epigenetic states and natural selection.
Here, the recent discoveries of plant siRNA and transposon regulations were reviewed. Both transposons and host plants have evolved ingenious strategies for survival. Clearly, transposons have acquired special relationships with host factors to proliferate within host genomes and have greatly affected genome evolution. On the other hand, plants have developed RNA-mediated interference (RNAi) that is a genome-based defense against transposons. The cause and effect of the survival strategies has played a significant role in shaping epigenetic regulation in plant development, natural variation, and stress responses. Transposons activated under environmental stress affecting neighbor gene expression may be a driving force of adaptation, while simultaneously creating a purging force to remove unfit variants if adjacent host genes become dysregulated due to transposon-derived siRNAs via the RdDM pathway. How these adaptation forces are balanced in nature remains poorly understood; however, we are now in a better position to reveal the relationship between siRNAs and transposon regulation in plant development and genome evolution.
I would like to thank Jon Reinders for reviewing the manuscript. This work was supported by JST, PRESTO, Grant-in-Aid for Scientific Research on Innovative Areas (23119501), Grant-in-Aid for Young Scientists (B) (23770034), and the Akiyama Life Science Foundation.