The meiotic cohesin subunit REC8 contributes to multigenic adaptive evolution of autopolyploid meiosis in Arabidopsis arenosa

Abstract

Genome duplication, which leads to polyploidy, poses challenges to the meiotic segregation of the now-multiple homologous chromosome copies. Genome scan data showed previously that adaptation to polyploid meiosis in autotetraploid Arabidopsis arenosa is likely multigenic, involving genes encoding interacting proteins. But what does this really mean? Functional follow-up studies to genome scans for multigenic traits remain rare in most systems, and thus many mysteries remain about the "functional architecture" of polygenic adaptations. Do different genes all contribute subtle and additive progression towards a fitness optimum, or are there more complex interactions? We previously showed that derived alleles of genes encoding two interacting meiotic axis proteins (ASY1 and ASY3) have additive functional consequences for meiotic adaptation. Here we study derived versus ancestral alleles of the meiotic cohesin subunit REC8, which has roles in chromatin condensation, recruiting the axes, and other critical functions in meiosis. We use genetic and cytological approaches to assess the functional effects of REC8 diploid versus tetraploid alleles, as well as their interaction with ancestral versus derived alleles of ASY1 and ASY3. We show that homozygotes for derived (tetraploid) REC8 alleles have significantly fewer unpaired univalents, a common problem in neotetraploids. Interactions with ASY1 and ASY3 are complex, with the genes in some cases affecting distinct traits, and additive or even antagonistic effects on others. These findings suggest that the road to meiotic adaptation in A. arenosa was perhaps neither straight nor smooth.

Fig 1. Analysis of REC8 TTTT, TxD and DDDD metaphase I cells.

(A) Key explaining the shapes of different metaphase I chromosomal configurations. Example images of each metaphase configuration (stained with DAPI) are shown (bottom, each box) alongside a stick interpretation, with chromatin shaded in grey (middle) and a cartoon indicating the predicted centromere (black circle) and CO (open circle with cross) position on recombining chromosomes. (B) Stacked bar chart showing the mean proportional frequency of different bivalent shapes in REC8 TTTT, TxD and DDDD metaphase I cells. (C) Plots showing the number of different chromosomal configurations per cell in REC8 TTTT, TxD and DDDD metaphase I cells. Dots indicate trait means and error bars 95% confidence intervals calculated from GLMM models. Significant between genotype p values are indicated: * p < 0.05.

Fig 2. Analysis of REC8 TTTT and DDDD late-pachytene cells.

(A) Example images of REC8 TTTT (top) and REC8 DDDD (bottom) late-pachytene cells imaged using 3D-SIM and labelled for ZYP1 (red), ASY1 (green), HEI10 (grey) and DAPI (blue). Maximum intensity projections of 3D images are presented. Scale bar = 5 μm. (B) Top row—plots showing the number of late-HEI10 foci (left), number of SPS sites (middle) and total SC length in μm (right) per cell in REC8 TTTT and DDDD late-pachytene cells. Bottom row—plots showing the predicted number of metaphase I bivalents (left), multivalents (middle) and univalents (right) per cell based on the relative position of late-HEI10 foci and SPS sites along pachytene chromosomes in REC8 TTTT and DDDD cells. Dots indicate trait means and error bars 95% confidence intervals calculated from GLMM models. (C) Cumulative frequency plots showing the distance of single late-HEI10 foci from the nearest chromosome end (left) and the distance between two late-HEI10 foci (right) in units of relative SC length in REC8 TTTT (blue) and REC8 DDDD (red) late-pachytene cells. (D) Frequency of different synaptic and crossover outcomes.

Fig 3. Analysis of RAA TTT and RAA DDD metaphase I cells.

(A) Stacked bar chart showing the mean proportional frequency of different bivalent shapes in RAA TTT and RAA DDD metaphase I cells. (B) Plots showing the number of different chromosomal configurations per cell in RAA TTT and RAA DDD metaphase I cells. Dots indicate trait means and error bars 95% confidence intervals calculated from GLMM models. Significant between genotype p values are indicated: * p < 0.05.

Fig 4. Analysis of RAA TTT and RAA DDD late-pachytene cells.

(A) Example images of RAA TTT (top) and RAA DDD (bottom) late-pachytene cells imaged using 3D-SIM and labelled for ZYP1 (red), ASY1 (green), HEI10 (grey) and DAPI (blue). Maximum intensity projections of 3D images are presented. Scale bar = 5 μm. (B) Top row—plots showing the number of late-HEI10 foci (left), number of SPS sites (middle) and total SC length in μm (right) per cell in RAA TTT and DDD late-pachytene cells. Bottom row—plots showing the predicted number of metaphase I bivalents (left), multivalents (middle) and univalents (right) per cell based on the relative position of late-HEI10 foci and SPS sites along pachytene chromosomes in RAA TTT and DDD cells. Dots indicate trait means and error bars 95% confidence intervals calculated from GLMM models. Significant between genotype p values are indicated: *** p < 0.0005. (C) Cumulative frequency plots showing the distance of single late-HEI10 foci from the nearest chromosome end (left) and the distance between two late-HEI10 foci (right) in units of relative SC length in RAA TTT (blue) and RAA DDD (red) late-pachytene cells. (D) Frequency of different synaptic and crossover outcomes.