Plant enhancers exhibit both cooperative and additive interactions among their functional elements

Abstract

Enhancers are cis-regulatory elements that shape gene expression in response to numerous developmental and environmental cues. In animals, several models have been proposed to explain how enhancers integrate the activity of multiple transcription factors. However, it remains largely unclear how plant enhancers integrate transcription factor activity. Here, we use Plant STARR-seq to characterize 3 light-responsive plant enhancers-AB80, Cab-1, and rbcS-E9-derived from genes associated with photosynthesis. Saturation mutagenesis revealed mutations, many of which clustered in short regions, that strongly reduced enhancer activity in the light, in the dark, or in both conditions. When tested in the light, these mutation-sensitive regions did not function on their own; rather, cooperative interactions with other such regions were required for full activity. Epistatic interactions occurred between mutations in adjacent mutation-sensitive regions, and the spacing and order of mutation-sensitive regions in synthetic enhancers affected enhancer activity. In contrast, when tested in the dark, mutation-sensitive regions acted independently and additively in conferring enhancer activity. Taken together, this work demonstrates that plant enhancers show evidence for both cooperative and additive interactions among their functional elements. This knowledge can be harnessed to design strong, condition-specific synthetic enhancers.

Figure 1.

Enhancers from photosynthesis genes show light-responsive activity. A and B) Full-length (FL) enhancers, as well as 169-bp long segments from their 5′ or 3′ end, of the pea (Pisum sativum) AB80 and rbcS-E9 genes and the wheat (Triticum aestivum) Cab-1 gene were cloned upstream of the 35S minimal promoter driving the expression of a barcoded GFP reporter gene (A). All constructs were pooled and the viral 35S enhancer was added as an internal control. The pooled enhancer library was subjected to Plant STARR-seq in Nicotiana benthamiana leaves with plants grown for 2 d in normal light/dark cycles (+ light) or completely in the dark (− light) prior to RNA extraction (B). Enhancer strength was normalized to a control construct without an enhancer (log₂ set to 0). C) Light-responsiveness (log₂[enhancer strength^light/enhancer strength^dark]) was determined for the indicated enhancer segments. D) Transgenic Arabidopsis (Arabidopsis thaliana) lines were generated with T-DNAs harboring a constitutively expressed luciferase (Luc) gene and a nanoluciferase (NanoLuc) gene under control of a 35S minimal promoter coupled to the 35S enhancer or the 3′ segments of the AB80, Cab-1, or rbcS-E9 enhancers. E) Nanoluciferase activity was measured in 5 T2 plants from these lines and normalized to the activity of luciferase. The NanoLuc/Luc ratio was normalized to a control construct without an enhancer (none; log₂ set to 0). F) The mean NanoLuc/Luc ratio was compared to the mean enhancer strength determined by STARR-seq. Pearson's R², Spearman's ρ, and number (n) of enhancers are indicated. A linear regression line is shown as a dashed line. Error bars represent the 95% confidence interval. Box plots in B, C, and E represent the median (center line), upper and lower quartiles (box limits), 1.5× interquartile range (whiskers), and outliers (points) for all corresponding barcodes (B and C) or plant lines (E) from 2 (B and C) or 3 (E) independent replicates. Numbers at the bottom of each box plot indicate the number of samples in each group.

Figure 2.

The AB80, Cab-1, and rbcS-E9 enhancers contain multiple mutation-sensitive regions. A to C) All possible single-nucleotide substitution, deletion, and insertion variants of the 5′ and 3′ segments of the AB80 (A), Cab-1 (B), and rbcS-E9 (C) enhancers were subjected to Plant STARR-seq in N. benthamiana plants grown in normal light/dark cycles (light) or completely in the dark (dark) for 2 d prior to RNA extraction. Enhancer strength was normalized to the wild-type variant (log₂ set to 0). A sliding average (window size = 6 bp) of the mean enhancer strength for all variants at a given position is shown. The shaded area indicates the region where the 5′ and 3′ segments overlap. Mutation-sensitive regions in the 3′ enhancer segments are indicated by shaded rectangles labeled a to e.

Figure 3.

Mutation-sensitive regions harbor transcription factor binding sites. A) Four to 5 mutation-sensitive regions (shaded rectangles; labeled a to e) were defined for the 3′ segments of the AB80, Cab-1, and rbcS-E9 enhancers. The mutational sensitivity plots are reproduced from Fig. 2. B to E) Sequence logo plots were generated from the enhancer strength in the light (B to D) or dark (E) of all possible single-nucleotide substitution variants within the indicated mutation-sensitive regions of the AB80 (B), Cab-1 (C), or rbcS-E9 (D and E) enhancers. The sequence of the wild-type enhancer and the position along it is shown on the x axis. Letters with dark colors in the logo plot represent wild-type bases. The sequence logos for each region were compared to known transcription factor binding motifs and significant matches are shown below the plots. F) For each transcription factor binding motif matching a sequence logo plots derived from the saturation mutagenesis data in the light (mutagenesis; see B to D) or identified by the motif-scanning approach (scanning; see Supplementary Fig. S7), the correlation (Pearson's r) between the strength of an enhancer variant and the score of how well the variant sequence matches this motif is plotted as points. The lines represent the average correlation for all motifs of a given enhancer.

Figure 4.

Circadian oscillation is robustly encoded in the AB80, Cab-1, and rbcS-E9 enhancers. A) All possible single-nucleotide variants of the AB80, Cab-1, and rbcS-E9 enhancers were subjected to Plant STARR-seq in N. benthamiana leaves. On the morning of the third day after transformation (ZT 0), the plants were shifted to constant light. Leaves were harvested for RNA extraction starting at mid-day (ZT 8) and in 6 h intervals (ZT 14, 20, 26, and 32) afterwards. B) A sine wave with a period of 24 h was fitted to the enhancer strength of a given variant across all sampled time points. The fitted line is plotted together with the measured data points for the wild-type enhancers. The equilibrium point of the curves was set to 0. The amplitude is shown as a 2-sided arrow at the time of highest enhancer strength (peak time). The goodness-of-fit (R²) is indicated. The shaded gray area represents the timing of the dark period if the plants had not been shifted to constant light. C and D) Histograms of the difference between the amplitude (C) and peak time (D) of each single-nucleotide variant relative to the wild-type enhancer. For comparison, the difference in enhancer strength at ZT 8 is also shown in C. Variants with a below average goodness-of-fit are grayed out in D. Only data for the 3′ enhancer segments is shown.

Figure 5.

Epistatic interactions between single-nucleotide deletions. A to C) Selected single-nucleotide deletion variants (A; Δ1, Δ2, Δ3, …) of the 3′ segment of the AB80, Cab-1, and rbcS-E9 enhancers and all possible combinations with 2 of these deletions (A; Δ1 + Δ2, Δ1 + Δ3, Δ2 + Δ3, …) were subjected to Plant STARR-seq in N. benthamiana plants grown in normal light/dark cycles (B) or completely in the dark (C) for 2 d prior to RNA extraction. For each pair of deletions, the expected enhancer strength based on the sum of the effects of the individual deletions (additive model) is plotted against the measured enhancer strength. The color of the points represents the distance between the 2 deletions in a pair.

Figure 6.

The number, spacing, and order of mutation-sensitive regions affect enhancer strength. A) Fragments of the AB80, Cab-1, and rbcS-E9 enhancers spanning 1 to 3 mutation-sensitive regions (shaded rectangles; labeled a to e, ab, abc, bc, de) as well as a control fragment (ctrl) from a mutation-insensitive region in Cab-1 and a shuffled version of the AB80 fragment d were ordered as oligonucleotides. These fragments were randomly combined to create synthetic enhancers with up to 3 fragments which were then subjected to Plant STARR-seq in N. benthamiana plants grown in normal light/dark cycles (light) or completely in the dark (dark) for 2 d prior to RNA extraction. The mutational sensitivity plots are reproduced from Fig. 2. B) Violin plots of the strength of the synthetic enhancers grouped by the number of contained fragments. C) For each enhancer fragment, the area under the curve (AUC) in the mutational sensitivity plots was calculated and plotted against the fragment's enhancer strength. AUCs in the dark or light for rbcS-E9 fragments c and d, respectively, are shown in A. Pearson's R², Spearman's ρ, and number (n) of enhancer fragments are indicated. A linear regression line is shown as a dashed line. D to G) Plots of the strength of enhancer fragments (D) or fragment combinations (separated by a + sign and shown in the order in which they appear in the construct; E) in 3 replicates (points) and the mean strength (lines). Enhancer strength was determined using N. benthamiana plants grown in the light (F) or dark (G) prior to RNA extraction. H) Violin plots of the difference in enhancer strength between synthetic enhancers harboring the same 2 enhancer fragments but in different order. The P-value from a 2-sided Wilcoxon rank-sum test comparing light and dark results is indicated (p). Violin plots in B and H represent the kernel density distribution and the box plots inside represent the median (center line), upper and lower quartiles, and 1.5× interquartile range (whiskers) for all corresponding synthetic enhancers. Numbers at the bottom of each violin indicate the number of elements in each group. Enhancer strength in B to G was normalized to a control construct without an enhancer (log₂ set to 0).

Figure 7.

Enhancer fragments can be used to build condition-specific synthetic enhancers. A) Plot of the strength of synthetic enhancers created by randomly combining up to 3 fragments derived from mutation-sensitive regions of the AB80, Cab-1, and rbcS-E9 enhancers (see Fig. 6A) as measured by Plant STARR-seq in the light or dark. The synthetic enhancers were grouped into 4 categories: inactive, log₂(enhancer strength) ≤ 1 in both conditions; constitutive, similar strength in both conditions; light-activated, at least 2-fold more active in the light; dark-activated, at least 2-fold more active in the dark. The number (n) of synthetic enhancers in each category is indicated. B) Dual-luciferase reporter constructs (see Fig. 1D) were created for 11 synthetic enhancers (syn1–11). Nanoluciferase activity was measured in at least 4 T2 plants from these lines and normalized to the activity of luciferase. The NanoLuc/Luc ratio was normalized to a control construct without an enhancer (none; log₂ set to 0). Box plots are as defined in Fig. 1E. C) The mean NanoLuc/Luc ratio was compared to the mean enhancer strength determined by STARR-seq. A linear regression line is shown as a dashed line. Error bars represent the 95% confidence interval. The constituent fragments of the synthetic enhancers are indicated with fragments separated by a + sign. The first letter indicates the enhancer from which the fragment is derived (A, AB80; C, Cab-1; R, rbcS-E9) and the lowercase letters represent the fragment name. D) A linear model was built to predict the strength of the synthetic enhancers based on the strength of the constituent individual fragments. Hexbin plots (color represents the count of points in each hexagon) of the correlation between the model's prediction and the measured data are shown. In C and D, Pearson's R², Spearman's ρ, and number (n) of synthetic enhancers are indicated.