The shift from outcrossing to selfing is common in flowering plants^{1, 2}, but the genomic consequences and the speed at which they emerge remain poorly understood. An excellent model for understanding the evolution of self fertilization is provided by Capsella（荠菜属） rubella, which became self compatible <200,000 years ago. We report a C. rubella reference genome sequence and compare RNA expression and polymorphism patterns between C. rubella and its outcrossing progenitor Capsella grandiflora. We found a clear shift in the expression of genes associated with flowering phenotypes, similar to that seen in Arabidopsis, in which self fertilization evolved about 1 million years ago. Comparisons of the two Capsella species showed evidence of rapid genome-wide relaxation of purifying selection in C. rubella without a concomitant change in transposable element abundance. Overall we document that the transition to selfing may be typified by parallel shifts in gene expression, along with a measurable reduction of purifying selection.

We shotgun sequenced the genome of the C. rubella reference line Monte Gargano (Italy) to 22× coverage using a combination of platforms (Online Methods, Supplementary Table 1 and Supplementary Note). For the final assembly of 134.8 Mb, covering all eight chromosomes, we used a genetic map with 768 markers¹¹ (Supplementary Figs. 1 and 2, Supplementary Tables 2,3,4,5 and Supplementary Note), Arabidopsis lyrata synteny (Supplementary Fig. 3) and BAC and fosmid paired-end link support (Supplementary Note and Supplementary Fig. 4). We predicted 28,447 transcripts from 26,521 protein-coding genes and 86 microRNA loci (Table 1 and Online Methods). We also conducted de novo genome assemblies from Illumina libraries for an outbred C. grandiflora accession and the close outgroup species Neslia paniculata (Supplementary Note, Supplementary Fig. 5 and Supplementary Tables 6,7,8,9).

Sequencing： Whole-genome shotgun sequencing of C. rubella was conducted using the Monte Gargano (Italy) reference strain. The majority of the sequencing reads were collected with standard Sanger sequencing protocols and Roche 454 XLR and Illumina GAIIx machines at the DoE JGI in Walnut Creek, California (JGI sequencing protocols; see URLs). One linear Roche 454 library (ten runs, 2.78 Gb), three 2.5-kb insert size paired libraries (three runs, 713.1 Mb), two 4-kb insert size paired libraries (four runs, 434.5 Mb), one 8-kb insert size library (three runs, 935.4 Mb) and one 10-kb insert size library (two runs, 515.4 Mb) were sequenced with standard XLR protocols.

Affiliations

Corresponding authors

• Detlef Weigel
• Stephen I Wright

http://www.nature.com/ng/journal/v45/n7/full/ng.2669.html#ref11