The Arabidopsis genus: An emerging model to elucidate the molecular basis of interspecific differences in transposable element activity.

Arabidopsis thaliana is a model plant species and its molecular dissection has greatly contributed to our understanding of the systems preventing genome invasion by transposable elements (TE). Recent advances suggest that A. thaliana may be more efficient than its congener A. lyrata at controlling TE expression and proliferation. The comparative analysis of TE transcription in A. thaliana and A. lyrata, which differ by 40% in genome size, may help understand how silencing mechanisms contribute to the evolution of transposition rate, an important factor controlling genome size variation in plants and animals.

The number of transposable elements (TEs) is known to vary greatly among plant genomes and evolves as the net outcome of three major factors: transposition activity, TE removal and population genetic processes determining the efficiency of purifying selection against new insertions.- In the genus Arabidopsis, the two species A. lyrata and A. thaliana differ by ~40% in genome size., As much as 56% of the A. lyrata genome sequence that cannot be aligned to A. thaliana encodes TEs or simple repeats. Differences in TE content therefore explain a significant fraction of genome size differences. Out-crossing species are expected to be more efficient in removing deleterious mutations than inbred species. 1As a consequence, new transposition variants with deleterious effects on plant fitness should segregate at lower frequency in the species that can best purify deleterious mutations, i.e., in the outbred A. lyrata. However, transposition variants are not segregating at lower frequency in A. lyrata., Therefore, population dynamics of deleterious mutations do not influence significantly the difference shown by A. thaliana and A. lyrata in TE content.

Instead, evidence is accumulating that the two species differ in the transcriptional control of TEs. In the model plant species A. thaliana, this control is increasingly well understood.- Silencing of TEs is mediated by small interfering RNAs (siRNAs) that guide the deposition of DNA and histone methylation marks on homologous DNA stretches.- These marks then act in cis- to repress expression. In A. thaliana, most TEs reside in pericentromeric chromosomal regions that are transcriptionally inactive., Under stressful conditions, silencing can be transiently suppressed,, making room for potential transpositions. Knocking-out distinct components of the transcriptional gene silencing system can also activate specific subsets of TEs., The analysis of small RNAs expressed in both species showed that A. thaliana appears to be able to repress TE transcription more efficiently than A. lyrata. The specificity of siRNAs for their target TEs is strongly associated with the efficacy of TE silencing in both A. lyrata and A. thaliana, but A. thaliana shows a 3-fold greater proportion of siRNAs mapping uniquely to a single TE copy. This finding agrees with the observation that a greater number of TEs are expressed in A. lyrata. In A. thaliana, genes located less than 2.5 kb away from TEs tend to be less expressed, suggesting that silencing TEs can entail negative consequences for plant fitness if protein-coding genes in the vicinity of TE insertions are partially silenced and cannot be expressed properly. This cost seems weaker in the congeneric species A. lyrata where TEs are more abundant and tend to be closer to expressed genes. Indeed, in this species, TE insertions are associated with a reduced expression of their closest protein-coding neighbor only if they are located less than 1.5 kb away. Interestingly, protein-coding genes showing greater expression of the A. lyrata allele in F1 interspecific hybrids confirm that the silencing effect of TE proximity is greater in A. thaliana.

Using a novel approach to monitor genome-wide variation in TE cis-regulation, He et al. in the de Meaux laboratory have demonstrated major interspecific differences in the cis-regulation of TE silencing. In F1 interspecific hybrids, monitoring of allele-specific expression at 1535 loci annotated as TEs in A. thaliana showed that as many as 47% of TE loci display allele-specific expression and thus differ in their cis-regulation. Interestingly, almost all differentially expressed TEs expressed the A. lyrata allele more than the A. thaliana allele. Allele-specific expression of TEs is not observed between A. thaliana accessions. He, de Meaux and collaborators confirmed on a sub-sample of 18 loci that the upregulation of A. lyrata alleles in F1 interspecific hybrids reflects parental differences in TE expression, and is not simply a result of TE upregulation induced by “genomic shock,” a phenomenon often observed in interspecific hybrids or synthetic Arabidopsis allopolyploids. A previous analysis based on RNA-sequencing data reported that 8% of A. lyrata TE loci are expressed, but the study of He et al. might be more sensitive as the approach is not affected by coverage issues that prevent the detection of lowly expressed genes. Importantly, this study further supports the hypothesis proposed by Hollister et al. that interspecific differences in TE regulation depend on cis-acting marks deposited by the epigenetic machinery. For ten of 11 TEs examined, H3K9me2 methylation, a silencing histone mark, was detected in A. thaliana but not in A. lyrata. In addition, differentially regulated TEs were significantly enriched among TEs controlled by the DNA methyl-transferase MET1, suggesting that allele-specific expression of TEs depends on CG methylation., Since A. lyrata TE alleles are systematically upregulated, it is possible that unknown changes in the epigenetic machinery of the two species create genome-wide differences in epigenetic marks, which are maintained in interspecific F1 hybrids and act in cis- to preferentially silence the A. thaliana allele. We suspect that interspecific differences in the epigenetic machinery are mostly quantitative: A. lyrata MET1 shares 90% amino acid identity with its A. thaliana ortholog (A.P., unpublished data) and appears functional. In addition, A. thaliana and A. lyrata seem to have the same repertoire of epigenetic marks,, although the A. lyrata epigenome has not yet been characterized. Further experiments are warranted to confirm this hypothesis. In fact, sequence differences in TE regulatory regions probably also contribute to methylation differences between TE alleles. The endosperm expressed-gene FWA, for example, shows interspecific variation in patterns of epigenetic silencing likely driven by the presence of repeat sequences in the promoter and differential TE regulation associates with nucleotide differences in TE upstream regions.

TE transcriptional upregulation is sometimes associated with increased transposition rates. Do the cis-regulatory differences observed for Arabidopsis TEs reflect a difference in transposition activity? The genomic sequence of A. lyrata revealed that recently inserted TEs were more abundant in A. lyrata than in A. thaliana.,, Therefore, the genome size difference between the two species is not only due to a massive removal of TEs in A. thaliana: TE amplification in A. lyrata also plays a role. He et al. found that TEs showing differences in cis-regulation do not show a comparatively higher number of copies. Using published age estimates, they observed that LTR retrotransposons differentially expressed in F1 hybrids are among the youngest TEs in the A. thaliana genome. This suggests that A. thaliana silences recent TE copies more effectively, although older copies are similarly regulated in both species, a finding in agreement with the interspecific differences in uniquely mapping siRNAs mentioned above. Nonetheless, TE age was estimated based on the divergence of the terminal repeats in LTR retrotransposons, which are identical at the time of insertion. We observed that age estimates are not congruent between species. Orthologous LTR retrotransposons, defined as elements flanked by orthologous neighbor genes, were estimated to be older in A. lyrata. Since orthologous TE insertions should be the same age, this questions the accuracy of LTR age estimation. Population parameters, which differ widely between outcrossing A. lyrata and selfing A. thaliana, probably influence the rate of LTR divergence, making interspecific TE age comparisons difficult. In addition, age estimates would be flawed if LTRs undergo gene conversion. To our knowledge, the role of this potential confounding factor has not been considered in these two species. The age distribution of LTR retrotransposons in the two species therefore needs to be re-examined in greater depth. The existence of inter-specific differences in transcriptional silencing of TEs is now clear but whether they have caused the widely different TE contents of the genomes of A. lyrata and A. thaliana remains to be demonstrated.

Interestingly, not only the host silencing system is evolving in the Arabidopsis genus: the elements themselves have evolved new functions that may also contribute to interspecific differences in TE proliferation. A groundbreaking study has demonstrated that orthologous families of TEs can evolve different transposition site preferences. Tsukahara et al. introduced Tal1, an A. lyrata member of the COPIA93/Evade TE family,, into the A. thaliana genome and monitored its activity. In A. lyrata, this TE recently proliferated in centromeric regions. In contrast, COPIA93 members in A. thaliana are found in low copy numbers and insert in chromosome arms. When introduced into A. thaliana, Tal1 is not efficiently silenced, and proliferates by integrating specifically in the centromeres. The rate of Tal1 transposition is magnified in a ddm1 mutant background, suggesting that Tal1 does not entirely escape the A. thaliana defense system. The release of COPIA93 silencing in A. thaliana ddm1 mutants leads to new integrations in chromosome arms but never in centromeres. Preferential insertion into centromeric regions seems to be an ancestral property of Tal1 and thus COPIA93 shows a modified insertion site preference in the A. thaliana lineage. This experiment provides an admirable demonstration that TEs have evolved new insertion abilities since the separation of A. lyrata and A. thaliana.

The rich molecular and genomic toolbox available in A. thaliana and the availability of the A. lyrata genome should facilitate the identification of molecular factors controlling both host and invader variation. Interspecific variation in the Arabidopsis genus therefore promises to bring novel insights into the mechanisms controlling the evolution of both “selfish DNA” and its control by the host’s genomic system. It also promises to shed light on a long-standing question: whether selfish DNA and the host defense system coevolve.