Cis-regulatory evolution spotlights species differences in the adaptive potential of gene expression plasticity

Abstract

Phenotypic plasticity is the variation in phenotype that a single genotype can produce in different environments and, as such, is an important component of individual fitness. However, whether the effect of new mutations, and hence evolution, depends on the direction of plasticity remains controversial. Here, we identify the cis-acting modifications that have reshaped gene expression in response to dehydration stress in three Arabidopsis species. Our study shows that the direction of effects of most cis-regulatory variants differentiating the response between A. thaliana and the sister species A. lyrata and A. halleri depends on the direction of pre-existing plasticity in gene expression. A comparison of the rate of cis-acting variant accumulation in each lineage indicates that the selective forces driving adaptive evolution in gene expression favors regulatory changes that magnify the stress response in A. lyrata. The evolutionary constraints measured on the amino-acid sequence of these genes support this interpretation. In contrast, regulatory changes that mitigate the plastic response to stress evolved more frequently in A. halleri. Our results demonstrate that pre-existing plasticity may be a stepping stone for adaptation, but its selective remodeling differs between lineages.

Fig. 1. Overview of the experimental setup and number of genes responding to stress response, the time point of maximal gene expression change.

A Overview of the experimental setup for the desiccation stress for the three Arabidopsis species and two F1 hybrids A. thaliana x A. lyrata and A. thaliana x A. halleri. In four independent trials, aerial parts of the plants were cut and left to dry out on absorbent paper for 0, 1.5, 3, 6, 12 and 24 h under growth conditions. Samples for transcriptome sequencing were flash frozen after the treatment and used for RNA extraction. B Definition of basal and plastic cis-regulatory changes. If basal expression differences between the parents (step 1) can be explained by a significant basal allelic ratio (step 2) in the F1, a basal cis-acting change is contributing to the change in gene expression. If the ratio of the expression response between the parents (step 1) due to the stress can be explained by the significant change in the allelic ratio in the F1s (step 2), a plastic cis-acting change is contributing to the change in slope. Details for the analytical pipeline in Supplementary Fig. 13. C Number of genes showing significant up- and down-regulation at each time point after initiation of the stress in each of the three species. A. lyrata has a higher number of up- and down-regulated genes at intermediate time points. D Number of genes reaching their maximum log-fold change at each time point in each of the three species confirms the stronger response in A. lyrata at intermediate points. E–G Strong pairwise correlation in gene expression fold change between 0 and 6 h into the desiccation stress for each species pairs: E A. thaliana vs A. lyrata (R = 0.84, p < 1e–321), F A. halleri vs A. lyrata (R = 0.85, p < 1e–321) and G A. halleri vs A. thaliana (R = 0.83, p < 1e–321). This indicates that the overall response to the stress is similar between the species and that the direction of transcriptional plasticity predates species divergence.

Fig. 2. Basal expression differences between species depend on the direction of plasticity in their common outgroup species.

A Sketch illustrating what ortho- or para-plastic changes in the derived species (A. halleri or A. lyrata) correspond to for up- and down-regulated genes, using A. thaliana as a proxy for the plasticity and gene expression level in their common ancestor. Black dot: ortho-plastic change, gray dot: para-plastic change. Ortho- and para-plastic changes in A. halleri (B) and A. lyrata (C). x-axis: ratio of gene expression after 6 h in A. thaliana, y-axis: ratio of basal expression in each species, compared to expression in A. thaliana after 6 h. Bars represent the number in each quadrant of the plot. D Number of ortho- (black) and para-plastic (gray) changes in A. halleri and A. thaliana. This terminology helps describe the evolutionary change in any given species, but we note that orthoplastic basal regulatory changes in one species are de facto paraplastic in the outgroup species and vice versa. E The number of genes for which a basal cis-regulatory change was detected in hybrids (significant log2ratio of A. halleri/A. thaliana or A. lyrata/A. thaliana among alleles in F1 hybrid and between the corresponding parental species) or no cis-regulatory change (parental difference are not associated with a bias in allele specific expression, assumed to be controlled in trans).

Fig. 3. Attenuated response to the stress compared to the plasticity observed in their outgroup species.

A Sketch illustrating what magnified or mitigated response to stress in the derived species (A. halleri or A. lyrata) correspond to for up- or down-regulated genes, using A. thaliana as a comparison. Black arrow: magnified response, gray arrow: mitigated response. Magnified and mitigated response in A. halleri (B) and A. lyrata (C). y-axis: ratio of the response ratios, x-axis: ratio of gene expression after 6 h in A. thaliana. Bars represent the number in each quadrant of the plot. D Number of genes with a magnified (black) and mitigated (gray) response in A. halleri and A. lyrata, compared to A. thaliana. E The number of genes that have a plastic cis-regulatory change (the ratio of the parental response is explained by the log2 ratios at xh vs. 0 h for A. halleri/A. thaliana or A. lyrata/A. thaliana in the F1 hybrid) or no cis-regulatory change (the ratio of the parental response is not associated with a bias in the allele specific expression, assumed to be controlled in trans).

Fig. 4. Distribution of derived cis-acting genes sets defined by their basal and/or plastic expression changes in A.halleri or A. lyrata compared to A. thaliana.

Ortho-Mag: orthoplasy and magnification, Ortho-Mit: orthoplasy and mitigation, Para-mit: paraplasy and mitigation, Para-Mag: paraplasy and magnification, Only-Mag: magnification only, Only-mit: Mitigation only, Only-Ortho: Orthoplasy only, Only-Para: Paraplasy only. A The number of observed (solid bars) and expected (hatched bars) genes in each of the 8 sets. Left: A. halleri; right: A. lyrata. “*”: Significant excess or depletion, inferred by a partial Chi square test with one degree of freedom. B Phylogenetic relationship between the three species allowed differentiating undetermined cis-acting changes (green arrow), which are shared between the two F1 hybrids, from the derived cis-acting changes, which are specific to one hybrid (blue and red arrow). C Proportion of derived and cis-acting modifications within all genes with a basal change (left) or a plastic response (right). Dashed line shows 50%, i.e. the expectation in the absence of a mutational bias. The total number of genes in these groups are shown in Supplementary Fig. 6. “*”: Significantly increased ot decreased number of derived changes in A. halleri and A. lyrata compared to changes predating the separation between the two species (basal: A. halleri vs ancestral p = 0.09959, A. lyrata vs ancestral p = 0.0002251; plastic: A. halleri vs ancestral p = 0.4068, A. lyrata vs ancestral p < 2.2e−16). D Proportion of derived and cis-acting modifications within each gene set. “*”: Significantly increased or decreased number of derived changes in A. halleri compared to A. lyrata. The association of basal and plastic changes in gene expression is not random (Chi-sq = 47.7 df = 7 p = 4.13e–08 and Chi-sq = 302.5 df = 7 p = 1.84e–61).

Fig. 5. Tajima’s D at synonymous and non-synonymous sites differ between 5 modes of plastictiy evolution.

Ortho-Mit: orthoplasy and mitigation, Only-Mag: magnification only, Only-mit: Mitigation only, Only-Ortho: Orthoplasy only, Only-Para: Paraplasy only. The plasticity categories Ortho-Mag, Para-Mit and Para-Mag were excluded due to having too few genes (see Fig. 4A) and therefore too high variance. A Ka/Ks ratio, B Ks between A. lyrata and A. thaliana, C Ka between A. lyrata and A. thaliana, D average π per gene, E Tajima’s D for synonymous sites and F nonsynonymous sites in A. lyrata. Significance between the groups was estimated based on pairwise comparison of 1000 bootstrap replicates, by using the union of the bootstrap values to calculate what proportion of the differences between the groups significantly differed from 0, multiplied by 2, to account for the fact that the test is one-sided. Box and whiskers depict the 75th and 95^th interquantile ranges, respectively, dots show outliers. No shared letters mean that the groups are different with a significance cut-off at p < 0.05. P-values of all pairwise comparisons can be found in Supplementary Data 8. G Relative change in mean Tajima’s D for non-synonymous sites between A. lyrata and A. thaliana. ((A. lyrata + 2) / (A. thaliana + 2)) and H Cumulative distribution function of the gamma distribution for the control and the 5 classes of plasticity. OrthoMag, ParaMit and ParaMag were removed due to the low number of genes in these plasticity classes.