ABSTRACT
Dioecy, the presence of male and female individuals, has evolved independently in multiple flowering plant lineages. Although theoretical models for the evolution of dioecy, such as the “two-mutation” model, are well established, little is known about the specific genes determining sex and their evolutionary history. Kiwifruit, a major tree crop consumed worldwide, is a dioecious species. In kiwifruit, we had previously identified a Y-encoded sex-determinant candidate gene acting as the suppressor of feminization (SuF), named Shy Girl (SyGI). Here, we identified a second Y-encoded sex-determinant that we named Friendly boy (FrBy), which exhibits strong expression in tapetal cells. Gene-editing and complementation analyses in Arabidopsis thaliana and Nicotiana tabacum indicated that FrBy acts for the maintenance of male (M) functions, independently of SyGI, and that these functions are conserved across angiosperm species. We further characterized the genomic architecture of the small (< 1 Mb) male specific region of the Y-chromosome (MSY), which harbors only two genes significantly expressed in developing gynoecia and androecia, respectively: SyGI and FrBy. Resequencing of the genome of a natural hermaphrodite kiwifruit revealed that this individual is genetically male but carries deletion(s) of parts of the Y-chromosome, including SyGI. Additionally, expression of FrBy in female kiwifruit resulted in hermaphrodite plants. These results clearly indicate that Y-encoded SyGI and FrBy act independently as the SuF and M factors in kiwifruit, respectively, and provide insight into the evolutionary path leading to a two-factor sex determination system but also a new breeding approach for dioecious species.
MAIN TEXT
In flowering plants, hermaphroditism is ancestral and most common, but a minority of plant species have evolved separate sexes (dioecy), in a lineage-specific manner (1–3). Similar to mammals, sexuality in plants is often determined by a heterogametic male system with XY chromosomes, where the Y chromosome is thought to carry one or two male-determining factors (1, 3, 4, 5). Previous analyses of Y chromosome evolution in plants, initially in Silene latifolia and Carica papaya, have revealed long non-recombining male-specific regions, which encompass many genes (6–8), although the sex-determining genes have not been fully characterized. Recently, Y chromosome-encoded sex determinants have been identified in persimmons (Diospyros spp.) and garden asparagus (Asparagus officinalis) (9, 10). In persimmons, the Y-encoded pseudogene OGI encodes a small-RNA targeting its autosomal counterpart gene, MeGI. To the best of our understanding, OGI is sufficient for expression of maleness and repression of female development (3, 9, 11). In garden asparagus, two Y-encoded factors, named SOFF and aspTDF, act independently to suppress gynoecium and promote androecium development, respectively (10). The asparagus observation is consistent with a previously proposed theoretical framework, called the “two-mutation model” (12, 13), while the persimmon case is not. This model proposes that evolution from an ancestral hermaphrodite could occur if females carried a mutated (non-functional) version of a male promoting factor (M) on the proto-X chromosome, resulting in establishment of gynodioecy. A second mutation, a gain-of function suppressor of feminization (SuF) on the proto-Y chromosome would then establish males. Together, these sex-determining mutations may, if closely linked, define a genome region resembling an XY chromosome pair or sex-linked genome region (13). Still, the evolutionary pathways governing the transitions into dioecy are poorly understood because only a few examples have been characterized to date.
Kiwifruit, a major fruit crop consumed worldwide, belongs to the genus Actinidia, in which most species are dioecious (14). The sexuality in kiwifruit is genetically controlled by a heterogametic male system (i.e., XY system). A Y-encoded cytokinin response regulator, named Shy Girl (SyGI), acts as one of the two putative sex determinants, the suppressor of female development (SuF) (15). The other sex determinant, the putative male promoting factor (M) has not been identified. Although the establishment of SuF in Actinidia is estimated to have occurred approximately 20-mya (15) and predated the divergence of the Actinidia species, kiwifruit still carries incipient and homomorphic sex chromosomes, including a small sex-determining region (15–17). Recent breeding has derived some hermaphroditic and neuter individuals in A. deliciosa, which constitute an additional resource for the identification of a second sex determinant in kiwifruit. Here, we attempted to identify the male promoting sex determinant (M factor), by fine assessment of the genes located on the male-specific region of the Y-chromosome (MSY), and identification of genes differentially expressed between male and female at an early stage of tapetum differentiation. The function of the M factor was validated in model plants and in kiwifruit, resulting in the first development of an artificial hermaphrodite crop from a dioecious individual. Finally, we further assessed the evolution of the two sex determinants on the Y-chromosome, unveiling transitions into and out of dioecy in kiwifruit.
Previously, genomic sequencing reads from F1 sibling trees derived from an interspecific cross, A. rufa sel. Fuchu × A. chinensis sel. FCM1 were used to identify and assemble the potential male specific region of the Y chromosome (MSY) of A. chinensis (15). The 249 resulting contigs, totaling approximately 0.5Mb in length, contained Y-specific sequences with perfect co-segregation with the plants’ sex, and included 61 hypothetical genes (15). In kiwifruit, androecia differentiation between males and females is observed during tapetum degeneration (stage 3-4 in Fig. S1) (15, 18, 19). To identify candidate male promoting (M) factors within the MSY, we conducted mRNA-Seq analyses on developing anthers (5 males and 5 females) before tapetum degeneration (“stage 1-2” in Fig. S1) from the F1 population described above. The mRNA-Seq reads were mapped to the 61 hypothetical candidate genes. Only one of them exhibited male-specific expression (RPKM > 1). This gene included a fasciclin domain, which are typically involved in cell adhesion (Table S1). This fasciclin-like gene was named “Friendly Boy (FrBy)”, as a potential counterpart of the SuF sex determinant, Shy Girl (SyGI). FrBy is nested within the monophyletic MTR1 family (Fig. 1a). In rice, MTR1 contributes to tapetum degradation via programmed cell death (PCD), resulting in male fertility (20). The kiwifruit FrBy and its orthologs in Nicotiana tabacum (FAS1 domain protein), Arabidopsis thaliana (AT1G30800) and rice (Oriza sativa, MTR1), showed no significant differentiation according to site-branch-specific evolutionary rate analysis against the other branches (Fig. 1b), suggesting conserved protein function. The presence of the FrBy gene was male-specific in a wide variety of Actinidia species (Fig. 1c). The expression of FrBy was specific to early developing androecia (Fig. 1d-e). Tapetum cell-specific qRT-PCR using laser capture microdissection (Supplemental Figure S2) and in situ RNA hybridization analysis (Fig. 1f-h) both indicated that FrBy expression in androecia was confined to tapetal cells in stage 1-2 and possibly to meiocyte or tetrads. This is consistent with previous observations of MTR1 in rice (Tan et al. 2012) and with its putative function to contribute to tapetum degradation following PCD in kiwifruit (Supplemental Fig. S1 and S3) (19). Differentially expressed genes (DEGs) in male and female anthers (Supplemental Table S2, Figure S4) were also consistent with the potential function of FrBy. We identified 538 DEGs (FDR < 0.1) when analyzing transcriptome data from developing anthers (as described above). In those 538 DEGs, GO terms involving PCD and phosphorylation signals were highly enriched (Supplemental Table S3). Not only PCD, but abundant phosphorylation signals are indispensable for proper tapetum maturing and degradation (21–23). Furthermore, an ortholog of Tapetal Development and Function 1 (TDF1) or MYB35, a key gene in tapetum maturation in Arabidopsis (Zhu et al. 2008) and one of the two sex determinants in dioecious garden asparagus (10, 24, 25), was detected as one of the male-biased DEGs in kiwifruit, although this gene (Acc30672.1) was not located within the MSY (Supplementary Table S2, Figure S5).
To investigate the function of FrBy, we first used the CRISPR/Cas9 gene editing system in two distantly related model plants, Arabidopsis thaliana and Nicotiana tabacum (Supplemental Table S4). Although the Arabidopsis genome includes three paralogs of FrBy, only one, AT1G30800, was in the cluster which was conserved across the Brassicaceae species (Fig. 1a). In Arabidopsis, the AT1G30800-null lines (Supplemental Figure S6) were self-sterile, with low pollen germination rates (Fig. 2g), but could successfully produce seed after being crossed to control male plants (Fig. 2a-e). The null line showed substantial delay in tapetal layer degradation (Supplemental Fig. S7), which is consistent with the development of female kiwifruit plants (Fig. S1) (18, 19). On the other hand, the lack of AT1G30800 had no significant effect on female reproductive function (P > 0.1, Supplemental Fig. S8). In N. tabacum, knock-out mutation of the FrBy ortholog, FAS1 (fas1) (Supplemental Figure S9) resulted in male sterility, with substantial reduction in pollen germination rate, and was accompanied by a delay in tapetum degradation. The other organs, including the gynoecium, showed no differentiation compared to the control plants (Fig. 2h-o). The transgenic N. tabacum lines expressing the kiwifruit SuF gene, SyGI, under the control of native promoter (pSyGI-SyGI) exhibited female-sterility (15). Reciprocal crossing using control plants, pSyGI-SyGI, and fas1 indicated that SyGI and FAS1 independently promote gynoecium and androecium development, respectively (Fig. 2p). Importantly, male function in a fas1 null line could be complemented by introduction of the kiwifruit FrBy under the control of its native promoter (Fig. 2q-r, Supplemental Figure S10), indicating that FrBy can act to maintain male fertility via proper tapetum degradation in N. tabacum. These results all suggest that FrBy is likely to be the male promoting factor, and that the two sex determining genes, SyGI and FrBy work independently for female and male fertility, respectively, in kiwifruit. Furthermore, our phylogenetic and evolutionary analyses indicated that the function of this fasciclin-like monophyletic gene is highly conserved across angiosperm species.
Reference genome sequences for kiwifruit have been assembled from female (2A+XX) cultivars (26, 27) but the Y chromosome of kiwifruit (or Actinidia spp.) has not been sequenced to date. Here, we constructed the whole genome reference sequence of a male cultivar, Soyu, which is one of the main pollinizers used in Japan. The sequences were assembled using 10X Genomics Supernova v1.2.2, which is based on a long haploblocking method suitable for assembly of highly heterozygous diploid genomes (28). Downstream genomic analysis for the Soyu cultivar was conducted on the “pseudohaploid” version of the whole genome assembly. The drafted genome sequence covered ca 710Mb which corresponds to 94% of the estimated size of kiwifruit genome (758 Mb) (Hopping, 1994, Huang 2013), with N50 = ca 318kb for scaffolds (Supplemental Table S5). The genomic short-read sequences were generated from large DNA molecules partitioned and barcoded using the Gel Bead in Emulsion (GEM) microfluidic method of 10X Genomics (28). Thus, the information present in the GEM barcodes anchored to the assembled contigs reflect their physical distance, in which proximal contig pairs often share the same GEM barcodes. GEM barcodes were extracted from read pairs and linkage information was assigned using a custom script. To construct longer scaffolds within the MSY, we first identified 9 scaffolds, which together spanned 1.43 Mb, and contained ca 87% of the Y-specific contigs previously assembled (15) (Supplemental Table S6). To organize these scaffolds relative to each other, we applied DelMapper (6), an approach that employed the traveling salesman problem (TSP), used in radiation hybrid mapping of mammalian chromosomes (29, 30) or deletion mapping of Y-chromosome in dioecious Silene latifolia (6). Using this method, 8 of the 9 scaffolds were successfully organized. In this new assembled super-scaffold, the two sex determinants, SyGI and FrBy, are located on adjacent scaffolds, at an estimated distance of ca 500kb (Fig. 3a, Supplemental Figure S11). Mapping of genomic reads from male and female individuals in the KE population (15) indicated that a ~800-kb region enriched in male-specific sequences (putative MSY) was located at the center of this assembled super-scaffold (Fig. 3a). The putative MSY includes the two sex determinants and highly repetitive sequences, which is consistent with the structure of MSY in other plants or animals (10, 31). The putative pseudoautosomal region (PAR) appears to be single copy and exhibited no substantial gender-bias, and mostly flanked the putative MSY, although some PAR-like sequences were also located inside the putative MSY (Fig. 3a). In this super-scaffold, 145 genes were predicted using AUGUSTUS (32), and 30 of these genes were fully male-specific (Fig. 3b-c, Supplemental Table S7). Of the 30 male-specific genes, only SyGI and FrBy were substantially expressed (RPKM > 1) in carpel and anther, respectively (Fig. 3d-e, Supplemental Table S8), based on transcriptome information from the KE population (Akagi et al. 2018 for carpel) (15). These data support the hypothesis that they are the two factors determining sexuality in kiwifruit.
We further corroborated the role of these two sex determinants in two ways. Breeding programs have generated a few hermaphrodite accessions in hexaploid A. deliciosa (33). We sequenced one of them, the KH-line (Supplementary Table S9), which was thought to be a Y-dependent (or possibly X-dependent) hermaphrodite (6A+XXXXXYh or 6A+XXXXXXh, Supplementary Figure S12). Mapping of the KH and control A. deliciosa [male cv. Matua (6A+XXXXXY)] genomic reads to the Y-chromosomal scaffolds described above (Fig. 3) demonstrated that the KH-line carries the FrBy gene, but not the SyGI gene, either through one or several long deletion(s), including the loss of SyGI, or through gain of FrBy on the X chromosome from recombination with the Y chromosome (Figure 4a, Supplementary Figure S13). Consistent with this result, SyGI could not be amplified in the KH-line, while FrBy could (Figure 4b). This suggested that loss of SyGI (the SuF) in the Y, or gain of FrBy in the X, resulted in a natural hermaphrodite line (Figure 4c). Next, we set out to develop hermaphrodite kiwifruit artificially, as well as to further validate the FrBy function. We introduced the FrBy ORF under the control of its native promoter (pFrBy-FrBy) into a “rapid flowering” A. chinensis female cv. Hort16A. In this line, the CENTRORADIALIS (CEN) genes have been truncated by gene-editing, resulting in lines that bypass the long juvenile phase (ca 3-4 years) and flower precociously (34). As anticipated, the pFrBy-FrBy lines, which flowered 4 months after regeneration, were hermaphroditic, exhibiting restored androecium function. These produced fruits including fertile seeds after self-pollination (Fig. 4d-h, Supplementary Table S10). Pollen tubes from the pFrBy-FrBy lines grew similarly to those from male accessions, in contrast to the control lines (Figure 4g-i, Supplementary Figure S14). These results clearly indicated that FrBy acts as the M factor in kiwifruit sex determination. They also provide valuable insight into new breeding approaches for dioecious species.
Taken together, our results are consistent with the following evolutionary path for the transition from hermaphroditism to dioecy in Actinidia, based on two tightly linked genes within a small MSY: loss-of-function of FrBy established a proto-X chromosome, while lineage-specific gain-of-function in SyGI (15) derived a dominant suppressor of gynoecium development, establishing a proto-Y chromosome (Figure 5). This evolutionary process and the predicted function of the determinants are consistent with the “two-mutation model” (12, 13). This proposed evolutionary history of SyGI and FrBy is also consistent with those of the two-locus type sex determinants in Asparagus or Phoenix (10, 35), although the specific function of these sex-determinants are different. The putative SuF genes, SOFF for Asparagus and LOG1-like for Phoenix, were established by lineage-specific gene duplication/translocation on the Y chromosome; while the putative M genes, MYB35 (TDF1) for Asparagus and CPY703/GPAT3 for Phoenix were lost from the X chromosomes. Within the order Ericales, kiwifruit (Actinidia) evolved these Y-encoded sex determinants while persimmons (Diospyros), evolved a single sex determinant on the Y chromosome, OGI, which encodes small-RNAs repressing an autosomal feminizing gene, MeGI (9, 36). Despite their different specific functions, SyGI in kiwifruit and OGI in persimmon both act as dominant suppressors and were both derived from lineage-specific duplications, suggesting evolutionary consistency in how the diverse sex determination systems have evolved in angiosperms.
AUTHOR CONTRIBUTION
TA, IK, and RT conceived the study. TA designed the experiments. TA, SMP, EV, SSS, MS, AF, MJD, TW, RR and CV conducted the experiments. TA, SMP, EV, SSS, MS, IMH, and AF analyzed the data, SMP, MAM, PD, ACA, KB, and IK initiated/bred and maintained the plant materials. TA, SMP, EV, and IMH drafted the manuscript. All authors approved the manuscript.
AUTHOR INFORMATION
All sequence data generated in the context of this manuscript has been deposited in the appropriate DDBJ database: Illumina reads for gDNAseq and mRNAseq in the Short Read Archives (SRA) database (SRA IDs), the genomic contig sets constructed with 10X Genomics reads were submitted to Genbank (IDs).
Reprints and permissions information is available at www.nature.com/reprints.
The authors declare no competing financial interests.
Correspondence and requests for materials should be addressed to Takashi Akagi (takashia{at}okayama-u.ac.jp).
ACKNOWLEDGEMENTS
We thank Dr. Luca Comai (UC Davis Dept. Plant Biology and Genome Center) for technical advice and bioinformatics support, Drs. Yusuke Kazama and Kotaro Ishii (Riken Institute) for technical support for using the DelMapper program, and Niels Nieuwenhuizen and Jane (Lei) Zhang for vector construction. The KE population were originally provided from Kagawa Prefectural Agricultural Experiment Station. Some of this work was performed at the Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley, supported by NIH S10 OD018174 Instrumentation Grant. This work was supported by PRESTO Grant Number JPMJPR15Q1 (to TA) and JPMJPR15Q6 (to SSS) from the Japan Science and Technology Agency (JST), by a Grant-in-Aid for Scientific Research on Innovative Areas No. J16H06471 (to TA) from JSPS, and by the National Science Foundation (NSF) IOS award under Grant No. 1457230 (to IMH)