Pervasive under-dominance in gene expression underlying emergent growth trajectories in Arabidopsis thaliana hybrids

Background Complex traits, such as growth and fitness, are typically controlled by a very large number of variants, which can interact in both additive and non-additive fashion. In an attempt to gauge the relative importance of both types of genetic interactions, we turn to hybrids, which provide a facile means for creating many novel allele combinations. Results We focus on the interaction between alleles of the same locus, i.e., dominance, and perform a transcriptomic study involving 141 random crosses between different accessions of the plant model species Arabidopsis thaliana. Additivity is rare, consistently observed for only about 300 genes enriched for roles in stress response and cell death. Regulatory rare-allele burden affects the expression level of these genes but does not correlate with F1 rosette size. Non-additive, dominant gene expression in F1 hybrids is much more common, with the vast majority of genes (over 90%) being expressed below the parental average. Unlike in the additive genes, regulatory rare-allele burden in the dominant gene set is strongly correlated with F1 rosette size, even though it only mildly covaries with the expression level of these genes. Conclusions Our study underscores under-dominance as the predominant gene action associated with emergence of rosette growth trajectories in the A. thaliana hybrid model. Our work lays the foundation for understanding molecular mechanisms and evolutionary forces that lead to dominance complementation of rare regulatory alleles. Supplementary Information The online version contains supplementary material available at 10.1186/s13059-023-03043-3.


Figure S4
. BTH treatment reduced rosette size in both inbreds and F 1 s.LMM spline fitting of cluster mean expression levels with 95% Bayesian credible intervals for dominant gene clusters not shown in Fig. 3. Mean expression level across all genes within a cluster for inbred parents (purple dots) and F1 hybrids (turquoise dots) against last day rosette area (mm 2 ) were plotted.Clusters 1-5 showed little rosette size-expression level association, while clusters 8-13 showed monotonic decrease of cluster mean expression level with increased plant size.While F1 hybrids exhibited the same trend as the inbred parents, the mean expression levels are consistently lower in F1 hybrids across the entire rosette size range for cluster 8-13.

Figure S5 .
Figure S5.Randomly selected genes do not show differential expression-size correlation in inbreds and hybrids.A. Heatmap of the average expression of gene clusters from 500 randomly sampled genes, arranged in the same order as in Fig. 2A.B. Randomly sampled genes, which show little expression-size covariation and no differences between inbreds and F1s.

Figure S7 .
Figure S7.Consistent and significant reduction in rosette area by BTH in both batches of the SHB2 experiment.Each tray was a treatment unit.Shown are boxplots of rosette area (mm 2 ) in each tray with mock (green) or BTH treatment (orange).Only trays with parent-hybrid trios further used for transcriptome analysis are included.The rosette area after treatment with BTH was reduced compared to mock treatment (2-waynested ANOVA, mock: 79.2 ± 3.0 mm 2 , BTH treatment: 39.5 ± 4.2 mm 2 ; p < 10 -16 ).

Figure S8 .
Figure S8.Reaction norm of rosette area (mm 2 ) after mock and BTH treatments for all inbred parents and F 1 hybrid trios.Solid lines connected the mean rosette area under the two treatments, with the dots illustrating individual rosette area of each plant and the error bars showing standard deviation of the biological replicates.

Figure
Figure.S9.BTH-responsive genes sorted into 61 clusters.Genes were sorted based on spline regression of rosette area MPH (mm 2 ) to their expression MPH across all trios.Shown are a graphic representation of the clusters.Each thin line represents a gene, and the thick line represents cluster mean (green: mock, orange: BTH).The 61 clusters were subsequently sorted into 12 general categories based on the regression trends in mock and BTH treatments.

Figure S10 .
Figure S10.Positive genes are enriched for genes encoding thykaloid-localized proteins that are involved in photosynthetic process.Fisher's Exact Test was used against the background list of filtered expressed genes in SHB2 dataset, using plant GOslim based on TAIR10 annotation.

Figure S11 .
Figure S11.Top 3 motif enrichment results for All-BTH negative genes.All-BTH negative genes are all genes showing negative correlations between dominance of expression with dominance in F1 rosette size when treated with BTH).No.Target: number of genes from All-BTH negative gene list carrying target motif in cis, No.Bkgd: number of genes from background gene list carrying target motif in cis.

Figure S12 .
Figure S12.Common additive genes.A. Venn diagram showing the overlapping of additive genes in SHB1 and SHB2 experiments.B. GO-term enrichment diagram of common additive genes.

Figure S13 .
Figure S13.Additive genes in SHB2.This gene set (n=901) is enriched for GO terms of response to stress and (biotic) stimuli, cell death, and secondary metabolic process.

Fig. S14 .
Fig. S14.Efficient induction of defense responses in A. thaliana accessions with the BTH dosage used.qRT-PCR of two defense marker genes, PR1 and NPR1, in 16 inbred accessions after BTH and SA treatment.Each dot represents the mean ΔΔCq value against housekeeping genes in mock-treated plants, with error bars indicating the standard deviation of the biological replicates.

Fig
Fig. S15.Genetic distance correlates poorly with absolute rosette size mid-parent heterosis.A-D.Absolute rosette size mid-parent heterosis (mm 2 ) regressed against inter-parental hamming distance (as fractions of total polymorphic SNPs) calculated using whole genome SNPs, or SNPs located within the B. additive genes, C. positive genes, and D. Negative genes.E. Kernel density plots of inter-parental hamming distance as the fraction of all filtered diallelic SNPs.Subsetting SNPs to given gene context did not dramatically change the hamming distance distribution.Curve in shaded grey shows the whole-genome hamming distance profile of all 1001 genome accessions.