The BRAHMA-associated SWI/SNF chromatin remodeling complex controls Arabidopsis seed quality and physiology

Abstract

The SWI/SNF (SWItch/Sucrose Non-Fermentable) chromatin remodeling complex is involved in various aspects of plant development and stress responses. Here, we investigated the role of BRM (BRAHMA), a core catalytic subunit of the SWI/SNF complex, in Arabidopsis thaliana seed biology. brm-3 seeds exhibited enlarged size, reduced yield, increased longevity, and enhanced secondary dormancy, but did not show changes in primary dormancy or salt tolerance. Some of these phenotypes depended on the expression of DOG1, a key regulator of seed dormancy, as they were restored in the brm-3 dog1-4 double mutant. Transcriptomic and metabolomic analyses revealed that BRM and DOG1 synergistically modulate the expression of numerous genes. Some of the changes observed in the brm-3 mutant, including increased glutathione levels, depended on a functional DOG1. We demonstrated that the BRM-containing chromatin remodeling complex directly controls secondary dormancy through DOG1 by binding and remodeling its 3' region, where the promoter of the long noncoding RNA asDOG1 is located. Our results suggest that BRM and DOG1 cooperate to control seed physiological properties and that BRM regulates DOG1 expression through asDOG1. This study reveals chromatin remodeling at the DOG1 locus as a molecular mechanism controlling the interplay between seed viability and dormancy.

Figure 1.

Seeds transcriptome and metabolome of brm and dog1 single and double mutants. A) Identification of DEGs in mature, dry brm-3, dog1-4, and brm-3dog1-4 seeds compared to Col-0 WT (3′RNA-Seq, differential analysis was performed using DESeq2—genes with FDR < 0.05 and absolute fold change > 1 were considered as differentially expressed). B) Analysis of overlap between genes whose expression was affected in analyzed mutants (extended graph version shown in Supplementary Fig. S1C). C) Self-clustering of expression profiles for genes differentially expressed in brm-3, identifies the DOG1-dependent genes among ones affected in brm mutant seeds. Number of DEGs is indicated on panels. D) Heatmap of genes’ expression for genes misregulated in brm-3 mutant across mutants used. Genes marked as DOG1 gene-dependent show suppression of the brm effect in the double brm-3dog1-4 mutant. Second column colors correspond to specific color of the cluster. E) Chemical enrichment analysis of brm and dog single and double mutants. Colored circles represent clusters of metabolites from given chemical families (red, increased cluster; blue, decreased cluster; purple, increased and decreased metabolites in a cluster). The number of metabolites as indicated as circle size. F) The graph represents the fold change of glutathione levels compared to Col-0 in the seeds of mutants indicated values were ranked into groups as indicated by the respective letter using a Student–Newman–Keuls test, n = 4.

Figure 2.

The brm-3 mutant seeds showed multiple morphological and physiological defects. A) Scanning electron visualization of seed from Col-0 WT, brm-3, dog1-4, and brm-3dog1-4 mutants. Bar corresponds to 200 nm. B) Seeds size analyzed using Boxed robot. C) Seed yield analyzed based on total number of seeds produced by mature plants. D) Seed longevity analyzed using artificial ageing. E) Germination in presence and absence of 100 mm NaCl. F) Primary seed dormancy analyzed with freshly harvested seeds. G) Secondary dormancy analysis for seed of brm-3, dog1-4, and brm-3dog1-4 mutants. Asterisks indicate significant differences compared to Col-0 dry seeds. Statistical analysis applies to all figure panels; t-test, *, P < 0.05, **, P < 0.01 and ***, P < 0.0001; n = 4, one biological replicate is a mixture of independent 5 plants; error bars represent standard deviation.

Figure 3.

BRM directly regulates DOG1 antisense transcription to control seed secondary dormancy. A) BRM ChIP-qPCR in dry seeds and B) seeds subjected to 3 days of secondary dormancy induction. Col-0 and BRM-GFP brm-1 seeds were analyzed using GFP antibodies. The x-axis shows beginning of amplicon relative to TSS, TSS = 0. Percent of input normalized to PP2A gene region. C) RT-qPCR analysis of DOG1 sense and D) antisense transcripts in Col-0, brm-3, and 3xbrd mutants during secondary dormancy induction; E) RT-qPCR analysis of reporter lines activity during secondary dormancy induction for p_senseDOG1-LUC and F)  p_ASDOG1-LUC lines in Col-0 and brm-3 background. G) RT-qPCR for endogenous and LUC-DOG1-deltaTATA line activity during secondary dormancy induction shows that inactivation of asDOG1 results in stronger induction of DOG1 during secondary dormancy induction. H) BRM, SWP73A, and BRD1 genes expression analysis during SD induction in Arabidopsis seeds. RT-qPCR analysis in C to H) is normalized using UBC21 gene. Statistical analysis applies to all figure panels, t-test, *, P < 0.05, **, P < 0.01 and ***, P < 0.0001; n = 4, one biological replicate is a mixture of independent 5 plants; error bars represent standard deviation.

Figure 4.

The brm3 mutant shows enhanced chromatin accessibility at DOG1 3′ end during secondary dormancy induction. A) FAIRE in Col-0 and brm-3 seeds on the 3rd day and B) 5th day of secondary dormancy induction. Chromatin accessibility at DOG1 shown as % recovery to noncrosslinked samples (UNFAIRE) and relative to PP2A. The x-axis shows beginning of amplicon relative to TSS, TSS = 0. C to F) RT-qPCR analysis of α-, β-, γ-, and δ-DOG1 mRNA splicing forms during secondary seed dormancy induction. Transcript level of short G) and long H) polyadenylated DOG1 mRNA forms The x-axis shows time/days of secondary dormancy induction. Statistical analysis applies to all figure panels; t-test, *, P < 0.05, **, P < 0.01 and ***, P < 0.0001; n = 4, one biological replicate is a mixture of independent 5 plants; error bars represent standard deviation.

Figure 5.

Model of the BRAHMA-associated SWI/SNF complex control of Arabidopsis seeds quality and physiology. The BRAHMA-associated SWI/SNF complex controls seed yield, seed size, and plant hormonal crosstalk to a large extend independently of DOG1. The BRAHMA controls longevity and secondary dormancy by controlling DOG1 expression through DOG1 antisense (in red color; dark gray arrows). The BRAHMA also either directly or through DOG1 antisense negatively controls DOG1 gene expression (gray arrow) by affecting its alternative splicing and alternative polyadenylation.