The persimmon genome reveals clues to the evolution of a lineage-specific sex determination system in plants

Abstract

Most angiosperms bear hermaphroditic flowers, but a few species have evolved outcrossing strategies, such as dioecy, the presence of separate male and female individuals. We previously investigated the mechanisms underlying dioecy in diploid persimmon (D. lotus) and found that male flowers are specified by repression of the autosomal gene MeGI by its paralog, the Y-encoded pseudo-gene OGI. This mechanism is thought to be lineage-specific, but its evolutionary path remains unknown. Here, we developed a full draft of the diploid persimmon genome (D. lotus), which revealed a lineage-specific whole-genome duplication event and provided information on the architecture of the Y chromosome. We also identified three paralogs, MeGI, OGI and newly identified Sister of MeGI (SiMeGI). Evolutionary analysis suggested that MeGI underwent adaptive evolution after the whole-genome duplication event. Transformation of tobacco plants with MeGI and SiMeGI revealed that MeGI specifically acquired a new function as a repressor of male organ development, while SiMeGI presumably maintained the original function. Later, a segmental duplication event spawned MeGI's regulator OGI on the Y-chromosome, completing the path leading to dioecy, and probably initiating the formation of the Y-chromosome. These findings exemplify how duplication events can provide flexible genetic material available to help respond to varying environments and provide interesting parallels for our understanding of the mechanisms underlying the transition into dieocy in plants.

Fig 1. Characterization of the draft persimmon genome.

a, Fifteen pseudomolecules with the genetically anchored contigs. Black, white and gray bars indicate the positions of the original contigs that were assembled in forward, reverse, or unknown direction respectively. b, Relative SNP density in the KK population. c, Relative density of repetitive sequences d, Relative gene density. e, Syntenic relationships within the persimmon genome.

Fig 2. Characterization of lineage-specific whole-genome duplication events.

a, Distribution of silent divergence rates between homologous gene pairs within the Diospyros, Actinidia, Solanum, and Vitis genomes. Diospyros shows a peak, indicated by an asterisk, at the same dS value as the Solanum triplication (T-tri), indicating the concurrent whole-genome duplication events. b, Comparison of the 4-fold degenerative transversion rates (4DTv) between the putative paralogous gene pairs, in the Diospyros, Actinidia, and Vitis genomes. Consistent with the distribution of dS values, a peak, which corresponds to Dd-α, was detected specifically in the persimmon genome, as indicated by an asterisk (*). In the Actinidia and Vitis genomes, peaks putatively corresponding the Ad-α/β and the hexaploidization-γ, were detected, as shown by the green and gray bands, respectively. c, Comparison of the dS values between the paralogous pairs in the Diospyros genome (orange), and the dS values between the orthologs in Diospyros and Actinidia (green), and in Diospyros and Vitis (purple). d, Estimated divergence time between Diospyros, Actinidia, and Vitis, with Arabidopsis as the outgroup. The concatenated sequences of 175 conserved genes across these species were used to determine divergence time, based on the previous estimated divergence of Actinidia and Vitis at 117MYA in the TIMETREE database (http://www.timetree.org). e, Summary of the lineage-specific WGD events in the asterids. The time scale is estimated from dS values and previous reports [–23]. K-Pg, Cretaceous-Paleogene boundary.

Fig 3. Fate of paleoduplicated genes in Diospyros.

a, Distribution of the pairwise dN/dS values in the Dd-α-derived paralogous gene pairs, from the alignment of the full ORF sequences. Most gene pairs are under purifying selection (dN/dS < 1.0), while only approximately 0.3% of the gene pairs (shown in red) exhibited neutral selection (N.S.) or weak positive selection (dN/dS ~ 1.0). b, model for the detection of the genes that underwent significant site-branch specific positive selection (posterior probability > 0.99 in Bayes Empirical Bayes method) after Dd-α, using Actinidia, Solanum, and Vitis as outgroups. c, Functional annotation of the 9 genes that underwent significant site-branch specific positive selection after Dd-α. MeGI is highlighted in gray. d, Inter-chromosomal collinearity between Chr. 4 and Chr. 13. Genes pairs showing significant similarity (<e^-100 in blastp) are linked (green lines). The segments surrounding SiMeGI and MeGI exhibit syntenic collinearity. e, Synteny analysis of chromosomes 4 and 13, based on gene order using SynMap (CoGe). The dotted rectangle highlights blocks of gene pairs with dS values ranging between approximately 0.5 and 0.9, including the MeGI-SiMeGI pair. The MeGI-SiMeGI syntenic region is indicated by a red circle, a more detailed figure is available in Supplementary S7 Fig. f, Microsynteny analysis of the genomic fragments including SiMeGI and MeGI, using promer in MUMmer. g-h, Comparison of the expression patterns of paralog pairs derived from the Dd-α event, focusing on the sex differentiation stages. The ratio of expression levels in male versus female developing flowers (g) and mature flowers (h) were compared in the paralogs putatively derived from the Dd-α WGD event. The ratios were expressed in log₁₀ scale. Approximately 10% of the gene pairs exhibited a statistically significant (P < 0.01, 2x2 Fisher’s exact test, orange circles) expression bias between the two paralogs (S3 Dataset), and 18.5% of the gene pairs (N = 242) showed >5-fold differences between the two paralogs.

Fig 4. Lineage-specific adaptive evolution of MeGI.

a, Phylogeny of the HD-Zip1 type homeodomain genes in the D. lotus genome. Only MeGI and SiMeGI were nested within the MeGI/Vrs1-clade with statistically significant support (100/100 and 74/100 for the divergence of MeGI/Vrs1 clade and MeGI/SiMeGI subclade in D. lotus, respectively). b, Divergence of the MeGI/SiMeGI-like orthologs in the asterids and evidence of strong positive selection immediately after the Dd-α WGD event in Diospyros species (colored in red). No significant positive selection was detected elsewhere in this tree. Pairwise dN/dS values within the current MeGI (0.095) and SiMeGI (0.22) sequences suggest that both genes have been functionally fixed. c, Branch-specific dN/dS rates sliding window analysis of MeGI/SiMeGI-like genes from various asterid species. MeGI specifically exhibits positive selection in the 5’ region (~0-170bp). The three asterisks indicate the positions of the positively selected sites according to the site-branch specific detection analysis performed using PAML. The position of the homeobox domain (HB) is indicated by the thick black line. d, Sliding window assessment of the pairwise dN/dS values in the current MeGI and SiMeGI alleles. All three of the positions positively selected in MeGI sites after the Dd-α WGD event (asterisks) are under stronger purifying selection in MeGI than in SiMeGI, consistent with a situation of an adaptive evolution utilizing the mutations positively selected after WGD.

Fig 5. Functional differentiation between MeGI and SiMeGI.

a-h, N. tabacum transgenic lines expressing either of MeGI or SiMeGI under the control of the 35S promoter. The lines expressing MeGI (a-c) showed rudimental anthers (a) which did not produce functional pollen grains (b), and severe dwarfism with chlorophyll starvation and narrow leaves (c, see S7 Fig for the detail). The lines expressing SiMeGI (d-f) developed regular anthers (d) which produced fertile pollen (e), and showed moderate dwarfism (f). pis: pistil, ra: rudimental anthers, an: anthers. g-h, Both MeGI- and SiMeGI-overexpressing lines were phenotypically different from the control plants transformed with empty vectors (cont), but the MeGI-expressing lines exhibited more severe departure from the WT controls for specific traits, such as leaves width (see S9 Fig). Bars indicate 5mm for a and d, 50mm for c, f, g, and h. i-j, expression patterns of MeGI, SiMeGI, and PI, with actin as a positive control, in the transgenic lines transformed with CaMV35S-MeGI (i) and CaMV35S-SiMeGI (j). k, DNA motifs identified as preferentially bound to following transcription factors all nested within the MeGI/SiMeGI clade: MeGI [25], SiMeGI (our experiments), and three Arabidopsis HD-ZIP1 genes [29], using DAP-Seq analyses (see Methods). l-n, expression patterns of MeGI and SiMeGI in buds and flower primordia were highly correlated (Pearson’s r > 0.7). Expression levels in female (l) and male (m) are expressed as RPKM values. n, Developmental stages.

Fig 6. Genomic context of the Y-chromosomal region surrounding OGI.

a, Read coverage from male (blue) and female (pink) samples and male/female coverage ratio across the scaffolds covering the male-specific region of the Y-chromosome. For both the male and female reads, expected coverage a single-copy sites is approximately 20 (grey lines across). This male-specific region was assembled via anchoring of the scaffolds with BAC sequences. Approximately 1.3Mb region was covered by Y-allelic scaffolds. More than 400kb of long repetitive sequences (dotted lines), flank OGI. Outer regions of these hyper repetitive sequences contain male-specific sequences (blue bands in M/F rate) and pseudo autosomal region (PAR)-like sequences (orange lines), where M/F rate was less than 70%, and the percentage of repetitive sequences was much lower. b, The silent divergence rate (dS) between X and Y alleles of the genes located in the PAR-like sequences (orange circles) decreases with distance to OGI. Stil, for most of these genes, the dS value between the X and Y alleles was larger than the average interspecific dS between the X alleles of D. lotus and D. mespiliformis (green square and dotted line), D. lotus and D. virginiana (blue square and dotted line), and D. lotus and D. kaki (red square dotted line). These results suggest that, in these PAR-like sequences, recombination between the X and Y alleles was suppressed before the divergence of Diospyros species, or at least predates the divergence between D. lotus and D. kaki. dS values for genes located in the regions closest to OGI are comparable to dS values between OGI and MeGI (gray circle, dS = 0.205), which suggest that little or no recombination occurred between these sequences after the establishment of OGI. In comparison, dS values between the X and Y alleles of genes located in the recombining region of chromosomes 15 are much lower (while circles on the right).

Fig 7. Hypothetical model for the role of duplication events in the evolution sexual systems in Diospyros.

The Dd-α event triggered positive selection on the 5-end and bZIP motifs, resulting in the acquisition of a new role for MeGI as repressor of male organs. This was potentially associated with the first switch in sexual system, from hermaprhodistism to monoeocy. The following duplication event, a segmental event, resulted in formation of OGI, containing an inverted repeat, which acquired the function of repression of MeGI expression via small-RNA production. This potentially triggered the establishment of the XY (heterogametic male) sexual system [5]. On the right, the dS scale, corresponds to the evolution of the MeGI/SiMeGI families and the observed sexual systems in each era. Based on the study of fossil records, unisexual (male) flowers were present during the Eocene era, which occurred significantly later than the Dd-α event (or the K-Pg boundary) [65]. The OGI/MeGI divergence and the establishment of the current function of OGI is ancestral to diversification within Diospyros and consistently, the Y-encoded OGI regulates dioecy in the whole Diospyros genus [5].