Recurrent evolution of heat-responsiveness in Brassicaceae COPIA elements

Transpositions and insertions of TEs may lead to loss of gene functionality [32, 33]. Therefore, TEs activity and mobility are tightly controlled by epigenetic means throughout the entire plant development [5, 6]. On the other hand, new insertions contribute to genome evolution and regulation of gene transcription [2]. Therefore, it was already suggested in the early days of transposon research that, under conditions when diversity of regulatory patterns in a population may provide a better basis for selection, limited transposon activation could be beneficial [34]. However, how occasional TE expression is provoked and how control is regained later is still a matter of debate. There is a rapidly increasing number of reports showing transient TE activation under various stress conditions reviewed in [1]. Therefore, it was hypothesized that stresses may open a window for transpositions. Here, we introduced the ONSEN (COPIA78) family as a model for understanding TE control and behavior under HS. ONSEN shows massive transcriptional upregulation upon HS in A. thaliana and new insertions in progenies of heat-stressed Pol IV mutant [7, 8, 17]. The molecular basis of ONSEN heat-responsiveness was puzzling until recently, when a typical HSFA2 TF binding HRE was identified in its cis-regulatory region [18]. Presence of canonical TF binding motifs in TE promoters was described for D. melanogaster and M. truncatula [15, 35]. However, the frequency of such activation strategy among TEs was unknown.

We analyzed LTRs of A. thaliana and A. lyrata heat-responsive COPIA TEs ONSEN, COPIA37, TERESTRA, and ROMANIAT5 for putative HREs. A minimum of three adjacent (5 bp) nGAAn motifs can form a basal HRE, whose activity will depend on their distance and the total number [20]. Heat-responsive COPIAs featured the whole spectrum of HREs ranging from the 4P types in ONSEN and TERESTRA, through 3P types in COPIA37 and ROMANIAT5 to a dozen of variable gap and step HREs in all these families. By comparing predicted HREs with transcriptional data, we conclude that gap and step HREs are mostly not sufficient to trigger HS-induced TE upregulation. This is congruent with their proposed low HSF binding efficiency [36]. Predicted 3P HREs correlated with up to a hundred-fold (COPIA37, ROMANIAT5) and 4P HREs with up to a thousand-fold (ONSEN, TERESTRA) transcript accumulation upon HS. This suggests a strong correlation between putative HREs and the transcriptional response of the TEs.

Previously it was shown that the TGS machinery antagonizes the TE activation [7, 17]. We found that the speed of re-silencing during or after HS depends on the HRE type. While ONSEN, with the strong 4P HRE, accumulated transcript during entire HS exposure, TEs carrying lower affinity HREs typically showed a maximum transcript amount at 6 h HS and lower levels at 12 h HS. This silencing can be reduced by treatment with DNA methylation inhibitors. Hence, stressed plants take active measures to prevent TE transpositions already during ongoing HS treatment. However, HS-induced TE activation must not always aim at transposition, but can be part of the plant regulome [2]. In A. thaliana, we found that heat-responsive AtROMANIAT5-2 controlled transcription of the APUM9 gene located downstream of the element. As we did not observe any evidence for high amount of a read-through transcript from ROMANIAT5-2 towards APUM9, we hypothesize that this transcriptional activation may be mediated rather by a specific three-dimensional chromatin organization at this locus. APUM9 gene was previously shown to be under control by HDA6 and synergistically by MOM1 and RdDM pathways, but not DDM1 and MET1 [30, 31]. Therefore, AtROMANIAT5-2 may represent a domesticated transposon with fine-tuned HS-regulated activation, contributing to transcriptional control of APUM9.

To challenge the hypothesis that HREs could be beneficial for TE amplification (but not necessarily for the host genome stability), we reconstructed evolutionary trajectories for HREs of ONSEN, COPIA37, TERESTRA, and ROMANIAT5 in the Brassicaceae. ONSEN was not heat-responsive in the early separated lineages represented by B. rapa and E. salsugineum, because its LTRs contained only one half of the 4P HRE (proto-HRE), which does not constitute a functional HRE. The proto-HRE became duplicated approximately 6–9 millions of years ago [25] and directly formed the present days 4P HRE found in the genus Arabidopsis and in the Australian species B. antipoda. Hence, ONSEN 4P HRE represents an evolutionary conserved cis-regulatory element. However, it should be noted that there are several other similarly or even more conserved regions within the ONSEN LTR. Whether they represent other TF binding sites and/or enhancers remains currently unknown. Furthermore, the ONSEN example shows that even high affinity HREs do not allow a TE to overrule the host genome defense, because their heat-responsiveness was lost in B. stricta, and the whole family became vanished from the C. rubella genome. In TERESTRA, high affinity 4P HREs evolved independently at two different LTR regions in the closely related species A. lyrata and B. stricta, while 3P HREs of COPIA37 emerged multiple times from a common nTTCn-rich LTR region. In contrast to ONSEN, HREs of these families are evolutionary young and species-specific. Whether they will be evolutionary successful, is an open question, but we speculate this to be the case for A. lyrata TERESTRA, where all genomic copies are full length, carry strong HRE, and respond to heat.

At present it is unknown whether higher temperatures in southern latitudes lead to greater amplification of heat-responsive TEs in subtropical relative to temperate zones. Although this is possible, there are also several factors that may act against such correlation. First, southern populations may reduce effects of HS by adaptation and growth at favorable microclimatic and/or temporal conditions [37]. Second, the genomes are subject to purification mechanisms and the higher transposition rate may be opposed by a greater frequency of TE removal [10]. Indeed, HS was shown to increase frequency of DNA sequence removal by a single strand annealing type of homologous recombination in transgenic constructs structurally resembling a LTR retrotransposon [38, 39]. Therefore, the final number of stress responsive TEs per genome may be the result of multiple effects acting in a complex network.