Pleiotropic effects of sex-determining genes in the evolution of dioecy in two plant species

Abstract

One reason for studying sex chromosomes of flowering plants is that they have often evolved separate sexes recently, and the genomes of dioecious species may not yet have evolved adaptations to their changes from the ancestral state. An unstudied question concerns the relative importance of such adaptation, versus the effects of the mutations that led to separate sexes in the first place. Theoretical models for such an evolutionary change make the prediction that the mutations that created males must have sexually antagonistic effects, not only abolishing female functions, but also increasing male functions relative to the ancestral functional hermaphrodites. It is important to test this critical assumption. Moreover, the involvement of sexual antagonism also implies that plant sex-determining genes may directly cause some of the sexual dimorphisms observed in dioecious plants. Sex-determining genes are starting to be uncovered in plants, including species in the genera Diospyros and Actinidia (families Ebenaceae and Actinidiaceae, respectively). Here, we describe transgenic experiments in which the effects of the very different male-determining genes of these two dioecious species were studied in a non-dioecious plant, Nicotiana tabacum. The results indeed support the critical assumption outlined above.

1. Introduction

In flowering plants, functional hermaphroditism is the most common sexual state. Such plants either have hermaphrodite flowers or are monoecious (with each individual having unisexual flowers of both sexes), and are called ‘cosexual’. However, 43 angiosperm families (175 genera) include species with separate sexes, and this ‘dioecious’ state is thought to have evolved independently from functionally hermaphroditic cosexual ancestors in different taxa [1,2]. The evolution of dioecy has been studied for a century, since the first discovery of genetic sex-determination in a plant [3], and sex chromosomes in Silene, Humulus, and Rumex, as reviewed by Westergaard in 1958 [4]. Theoretical models have suggested that the evolution of dioecy requires two mutations, one generating females by a mutated (non-functional) version of a male-promoting factor, M (M → m mutation), and one creating males.

For the femaleness (m) mutation to spread in a population and produce a substantial frequency of females, it must increase fitness, through producing outcrossed progeny (which is advantageous if inbreeding depression is severe), and/or re-allocating reproductive resources from male to female functions, increasing seed output or seed quality, compared with the ancestral state. Then, to establish males, a gain-of-function mutation is required, to wholly or partially suppress female functions (and therefore called an SuF factor) [5]. This model receives support from the documented inbreeding depression in many plants, and the observation that, in plant genera, the evolution of dioecy is correlated with an absence of self-incompatibility, suggesting that the ancestors of dioecious plants were often able to self-fertilize [6]. The two-mutation model is further directly supported in Silene latifolia (Caryophyllaceae). M and SuF factors can be separately deleted from this plant's Y chromosome [4], and have been mapped to distinct regions of the large Y-linked male-specific, or male-specific Y (MSY), region, using molecular markers [7]. Moreover, the model predicts that natural selection will favour closer linkage between the two genes with the mutations, because the advantage of the SuF mutation in increasing male fertility (and converting hermaphrodite individuals into more male-like forms, or complete males) is predicted to occur at the expense of female functions [5]. This model is therefore consistent with the evolution of sex-linked multi-gene genome regions in S. latifolia and papaya (Carica papaya, Caricaceae, see [8,9]) and heteromorphic, fully non-recombining sex chromosomal regions in some other dioecious plants [4].

However, a critical aspect of such models remains to be tested: the female-suppressing factor (SuF mutation) involved in the change from cosexuality to dioecy must simultaneously enhance male functions, as originally suggested by Darwin [10]. A reduction in female functions, producing male, or largely male, individuals, is not sufficient. This effect must be accompanied by increased male functions, i.e. a special form of pleiotropy, with sexually antagonistic effects, is required [5]. This prediction applies for SuF mutations that completely abolish female functions, or ones making the ancestral functional hermaphrodites quantitatively more male-like (creating ‘sub-dioecious’ populations, with so-called ‘inconstant males’, which can sometimes produce seeds under favourable conditions, whose evolution probably involves re-allocation of reproductive resources from female to male functions).

Without knowing the sex-determining genes in actual dioecious plants, this cannot be tested. The S. latifolia sex-determining genes have not yet been identified, but the recent identification of the sex-determining genes in several plants, the persimmon (Diospyros spp.) [11], garden asparagus (Asparagus officinalis) [12,13] and kiwifruit (Actinidia spp.) [14,15], now makes it possible to test whether the hypothesized effect enhancing male functions actually occurs when the female-suppressing factors of dioecious species are introduced into non-dioecious plants. We used the persimmon and kiwifruit in such tests. Both are members of long-lived dioecious genera, Diospyros and Actinidia, in the order Ebenales. Even though dioecy did not evolve recently in these species, their physically small Y-linked, or MSY regions allowed the identification of the sex-determining genes [11,14,15].

Our approach used transformation experiments of a model plant with hermaphrodite flowers, Nicotiana tabacum, to test pleiotropic effects in a species that has not adapted to unisexuality. To test whether the mutations that suppressed female functions in the actual evolution of dioecy satisfy the condition predicted above for SuF factors, we investigated the effects of SuF factors in Diospyros and Actinidia. It cannot be assumed that these mutations increase male functions as many mutations have similar effects in both sexes [16]. Their effect requires testing, and a specific goal of our study was to do this. Below, we discuss evidence arguing against the possibility that the development changes observed are related to the constitutive expression of a transgene.

Despite major differences in the two dioecious systems, outlined in the next two sections, both involve mutations in two genes. The prediction concerning the pleiotropic effects of the SuF mutation applies to both, and both systems are thus suitable for testing whether loss of female functions (caused by the respective SuF factors) is accompanied by increased male functions, i.e. whether sexually antagonistic (SA) effects are involved in the evolution of dioecy from functional hermaphroditism. Our experiments were designed to test:

• (i)whether the second, SuF, or maleness, mutation that decreases female functions, to create males, also increases male functions. The assumption that the SuF mutation shows a trade-off between the effects on male and female functions is crucial in theories for the evolution of dioecy; and
• (ii)whether the sex-determining genes have pleiotropic effects causing sexual dimorphisms (as opposed to the observed secondary sex differences being the result of changes that evolved after dioecy was established).

We also consider a third question, whether the female-determining mutation in persimmon was likely to have been advantageous when it arose (specifically whether the M → m male-sterility mutation that created females by abolishing male functions was also free from pleiotropic effects likely to decrease other components of fitness, including female functions).

2. Material and methods

(a). Plant resources and characterization of flowers

The plant materials are held (along with replicate DNA samples) in public repositories accessible for any research purposes, as detailed below. The genus Actinidia male and female individuals of the KE population derived from A. rufa sel. rFuchu × A. chinensis sel. FCM1 [14], were grown in the experimental orchard at Kagawa University, Kagawa, Japan (34.2°N, 134.0°E). The FCM1 and Fuchu selections, and the KE population, are held in the repository at the Kagawa Prefectural Agricultural Experimental Station (Fuchu 6117-1, Sakaide, Kagawa, 762-0024). Flowering branches or photographs of flower buds of five Actinidia species, A. chinensis, A. rufa, A. deliciosa, A. polygama and A. arguta, were gifts from Kagawa University, and the National Clonal Germplasm Repository—Tree Fruit & Nut Crops & Grapes: Davis , United States Department of Agriculture (USDA) The Agricultural Research Service (ARS).

For the genus Diospyros, male and female individuals of the KK population derived from crossing D. lotus cv. Kunsenshi-female × cv. Kunsenshi-male [11], were grown in the experimental orchard at Kyoto University, Kyoto, Japan (N34.7°, E135.8°). The KK population is held at the Kyoto University repository (Kitashirakawa, Oiwakecho, Sakyo-ku, Kyoto, 606-8502). Flowering branches of D. dygina and D. discolor were gifts from the National Clonal Germplasm Repository—Tree Fruit & Nut Crops & Grapes: Davis, Florida and Hawaii (USDA-ARS). The photographs of D. lotus and D. kaki flowers were taken in the Kyoto University experimental orchard. Flowering (fruiting) branches of Euclea pseudebenus were gifts from the Los Angeles County Arboretum.

The flower numbers per inflorescence were counted in each of 22 male and female individuals of the KE population [14], and in each of 35 male and female individuals of the KK population [11].

(b). Transformation of Nicotiana tabacum

In Actinidia, the sex-determinant with SuF male-determining function, Shy Girl (SyGI) [14], defines the Y-linked region. Its full-length genomic sequence, including its 1.5-kb 5′ promoter region, was amplified by polymerase chain reaction (PCR) with PrimeSTAR GXL (TaKaRa) from genomic DNA (gDNA) of A. chinensis cv. FCM1, using the following primers: SyGI-prom1500-1F-LIC2 (5′-TTCTAGTTGGAATGGGTTACCACCAAACAACAATAAATATTAGAGAG-3′) and SyGI-commonRR-spR-LIC2 (5′-TCCTTATGGAGTTGGGTTTCAATATTGGTACTTGATGTTGAGTTGG-3′), and to connect to the pPLV2 vector [17]. Full length genomic sequences of the Diospyros Y-linked sex-determinant, OGI, which acts in small-RNA production [11], and its autosomal target gene, MeGI [11], including their 1.5 kb 5′ promoter regions, were amplified from gDNA of a D. lotus cv. Kunsenshi-male, using the same method, and the following primers: NatMeGI-LIC2-stF-Gib (5′-GAATTCTAGTTGGAATGGGTTTTGTAATTTCGACCTGCACTCTCTAC-3′) and NatMeGI-LIC2-spR-Gib (5′-TCCTTATGGAGTTGGGTTTGTGCGAGAGAAGCCTAATGTAATT-3′) for MeGI, and OGI-long-stF-LIC (5′-GAATTCTAGTTGGAATGGGTCTAGATAAGAGAAATCCAAAAGACATATG-3′) and OGI-long-stR-LIC (5′-TGCAGTATGGAGTTGGGTTTCCTAGTGAGATGTCAAATTCCACTG-3′) for OGI, to connect to the pPLV2 and pPLV3 vector, respectively. We made the pPLV2-SyGI (pSyGI-SyGI), pPLV2-MeGI (pMeGI-MeGI) and pPLV3-OGI (pOGI-OGI) vector using the In-Fusion HD Cloning kit (Clontech, Tokyo, Japan) with pPLV2 or pPLV3 [17] digested by HpaI, according to the manufacturer's recommendations and previous reports [11,14].

Tobacco plants (N. tabacum) cv. Petit Havana SR1 were grown in vitro under white light with 16 h L : 8 h D cycles at 25°C until transformation. The binary construct was introduced into the A. tumefaciens strain EHA105, with the helper vector pSOUP [17], by electroporation. Young petioles and leaves of tobacco plants were transformed by the leaf disc method as previously described [11]. Transgenic plants were selected on Murashige and Skoog (MS) medium supplemented with 100 µg ml⁻¹ kanamycin or 50 µg ml⁻¹ hygromycin. Nicotiana tabacum FAS1a/b null lines that lack the orthologue of the Actinidia Y-linked male-determinant (or M factor), Friendly Boy (FrBy), and show male-sterility [15], were also created by self-fertilizing the Ntnull #3-FrBy 1 line, where ectopic expression of Actinidia FrBy had restored male fertility [15].

(c). Evaluation of phenotype in Nicotiana tabacum transgenic lines

Tansgenic lines showing substantial expression of the foreign genes in young shoots were assessed for phenotypic traits. Expression of each of the genes, SyGI, MeGI and OGI, was detected by PCR, as previously described [11,14]. For the pMeGI-MeGI + pOGI-OGI transgenic lines, we used individuals with significantly reduced MeGI expression, relative to pMeGI-MeGI-induced lines, for phenotypic assessment.

For each construct, at least 42 independent transgenic lines were regenerated, and of these, 15 transgenic lines were selected to derive T₂ generations for phenotypic assessment. Immediately after seed germination, transgenic plants were transferred from MS medium to culturing soil (Tsuchitaro, Sumirin Agro-products, Aichi, Japan). The plants were grown under white light (400–750 nm) with 16 h L : 8 h D cycles at 23°C. Flowering time in this study was measured from the time the plants were transferred into soil to the time of first flowering. For flower numbers, we counted the numbers in the first inflorescence. Pollen tube germination was assessed 3 h after placing the pollen grains on 15% sucrose/0.005% boric acid/1.0% agarose media at 25°C. The pollen germination rate was counted as the average percentages in batches of approximately 200 pollen grains from the first three flowers produced. Pollen grains exhibiting pollen tubes longer than the length of the grain were considered to have germinated.

3. Results and discussion

(a). Genetic sexual dimorphisms in Actinidia owing to pleiotropic effects of Shy Girl

The kiwifruit (and asparagus) systems resemble Westergaard's two-mutation model outlined in the Introduction, with females homozygous for a mutated (non-functional) version of a male promoting factor (M, called Friendly Boy in kiwifruit, and denoted by FrBy, and aspTDF in asparagus), and males carrying a suppressor of femaleness, or SuF, mutation (called Shy Girl and SOFF, respectively) [7,12–15]. In both species, the two factors are closely linked, but, unlike the situation in S. latifolia, the Y-linked regions are physically small.

The two cultivated kiwifruit species (A. deliciosa and A. chinensis) both show higher flower numbers per inflorescence in males than females (figure 1a–e); females generally have either a single flower per inflorescence, or two, and inflorescence with more than two flowers are rare (figure 1f). This difference is controlled by the sex-determining region, not the genetic background, because the males in a segregating F₁ population made from crossing A. rufa and A. chinensis ([8], see Methods section) commonly had more than three flowers per inflorescence, significantly more than in females (Mann–Whitney U-test p = 2.0 × 10⁻⁷). Flowering was also earlier in males than females in segregating F₁ populations of A. deliciosa, although this was not formally analysed [18]. Increased flower number and earlier flowering are therefore associated with having the male genotype, and these traits are shared by males of different species in the genus Actinidia, including A. arguta, probably the most distant relative of cultivated kiwifruit (figure 1g–n).

Figure 1.

Sexual dimorphisms in the genus Actinidia. (a,b) Representative male (a) and female (b) flowers, in a F₁ segregating population (KE) derived from A. rufa × A. chinensis [13]. An, anther; Pe, petal; Ca, carpel; DA, defective anther. (c,d) Representative architectures of male (c) and female (d) inflorescences in the same F₁ population, and in several Actinidia species (parts g–n, with triangles indicating the structures showing differences between the sexes). (e) Box plot of the numbers of flowers per inflorescence in 22 male and 24 female individuals from the same F₁ population. (f) Distribution of flower numbers with male and female inflorescence architectures in male and female individuals of the KE population. (g–n) show representative inflorescences architectures in male and female plants in A. chinensis (g,h), A. rufa (i,j), A. hypoleuca (k,l) and A. arguta (m,n). (Online version in colour.)

We tested the effects of the Actinidia Y-linked SuF factor, Shy Girl (SyGI) [14] by transforming N. tabacum plants with Shy Girl under the control of the native Actinidia promoter (pSyGI-SyGI). The female-suppressing effect (defective gynoecia) was expressed in N. tabacum, suggesting that this gene's action is similar in both species' flower developmental systems (figure 2a–e), confirming previous suggestive results [14]. Strikingly, the transformed lines had substantially increased bud and flower numbers, the predicted pleiotropic effect for an SuF factor that should benefit male functions (figure 2f–m; electronic supplementary material, table S1). On the other hand, N. tabacum plants that lack the orthologue of the Actinidia Y-linked M factor, Friendly Boy (the FrBy orthologue null genotype explained in the Methods section), showed no substantial changes in inflorescences or flowering time (figure 2m; electronic supplementary material, table S1).

Figure 2.

Ectopic effects of the Shy Girl in Nicotiana tabacum. (a–d) Transgenic lines transformed with pSyGI-SyGI (a) show defective carpels (labelled DC in parts a and e) and female-sterility (An, anther; Pe, petal). The control plants transformed with empty vectors (labelled cont in parts e and f) showed no phenotypic changes in the flowers (b) and bear fertile fruits (labelled Fr in part d). By contrast, the pSyGI-SyGI individuals produced no fertile fruits in self/cross-pollinations (c). They also had short pistils and their stigmas remained immature in comparison to the control lines (part e: Pi, pistil; Sg, stigma). (f–m) The numbers of flowers per inflorescence were substantially increased in the pSyGI-SyGI lines, starting at the flower buds initiation stage (f). (g, i and k) show the results for the pSyGI-SyGI lines, and (h, j and l) for controls. (m) shows a statistical test for flower numbers in 15 pSyGI-SyGI lines, 15 gene-edited lines in which the orthologue of the kiwifruit M factor, FrBy, was lacking, and 15 controls. Only the pSyGI-SyGI lines showed significantly increased flower numbers per inflorescence (electronic supplementary material, table S1). (Online version in colour.)

No other sexual dimorphism displayed by kiwifruit was detected in N. tabacum transformants. Flowering tended to be earlier than in control plants, resembling the kiwifruit sex difference, but not significantly (electronic supplementary material, figures S1A and S2 and table S1). No anther size changes were detected, in either direction (p > 0.1, electronic supplementary material, table S1), consistent with the expectation that a naturally evolved maleness factor should not cause male-deleterious effects, and no flower size differences were detected (electronic supplementary material, figure S1B,C), which is also consistent with kiwifruit male flowers. However, differences that are not natural sexual dimorphisms in kiwifruit were also seen: the transgenic plants' architecture, particularly internode length and numbers of axillary buds bursting, were often affected (electronic supplementary material, figure S3), and leaf shape was often narrow, as previously noted [14] (electronic supplementary material, figure S3 and table S1).

Given that Shy Girl protein is a general repressor of cytokinin signalling, a pathway required for proliferation of plant meristem cells in their undifferentiated state [14], kiwifruit maleness mutations might be expected to lead to early differentiation of reproductive structures, or precocious flowering, which, as just mentioned, was not clearly seen. However, the mechanisms regulating flowering time probably differ greatly in Actinidia and N. tabacum, as the former are perennial plants with a juvenile (non-reproductive) phase of at least 3 years, while the latter is an annual plant with a very short juvenile phase. Gene(s) other than SuF could thus be the cause of the earlier flowering of male than female kiwifruit. The sex difference might have evolved after dioecy evolved in a dioecious ancestor, via changes in gene(s) linked to the sex-determining locus, such as the Y chromosome-specific FT-like gene [14,15], or in unlinked genes that affect flowering time in a sex-specific manner.

(b). Genetic sexual dimorphisms controlled by the Diospyros OGI/MeGI system

The persimmon system has fundamentally different properties. The maleness gene, defining a Y-linked region, is named OGI, and is absent in females. OGI is a duplicated copy of another gene, MeGI, and causes maleness by encoding a small-RNA that inhibits MeGI's action; in the absence of OGI, MeGI promotes the gynoecium and represses the androecium, causing femaleness [11,19–21]. Females probably arose first, through a mutation in the MeGI gene, followed by the OGI duplication creating males [11]. A major difference from Westergaard's model is that the properties of the two genes enable the single factor OGI, with a presence/absence polymorphism (present in males and absent in females) to have bifunctional control of sex-determination, acting as both an SuF factor and an M factor. The spread of OGI therefore allows MeGI to spread throughout the species, becoming a non-polymorphic autosomal gene [11,19,22]. In this system, therefore, two mutations occurred during the evolution of dioecy, but they are unlinked, and no large Y-linked non-recombining region should evolve unless subsequent adaptations to unisexuality involve sexually antagonistic polymorphisms maintained in the nearby genome region. Whether a non-recombining region evolves, or not, will depend on whether such a polymorphism evolves [23].

The inflorescences in D. lotus are also sexually dimorphic. As in kiwifruit, persimmon females have a single flower per inflorescence, while males normally have a cyme-like inflorescence with three or more flowers (figure 3a–c). This dimorphism also co-segregates with the sex-determining locus genotype in two segregating F₁ populations, one derived from the cross of cv. Kunsenshi female × cv. Kunsenshi male, and the other from a cross with a different cultivar as the female parent, cv. Budougaki × cv. Kunsenshi male (figure 3d, Mann–Whitney U-test p = 1.1 × 10⁻¹⁵). Other Diospyros species display similar sexual dimorphism, including D. kaki (cultivated Oriental persimmon), D. digyna and D. discolor, and it is also observed in another genus in the family Ebenaceae, Euclea (figure 3e–l). In the persimmon F₁ populations just mentioned, males also tend to have a shorter juvenile phase and to flower precociously [24].

Figure 3.

Sexual dimorphisms in the genus Diospyros. (a,b) Appearance of representative female (a) and male (b) flowers, in a D. lotus F₁ segregating population, named KK [10]. An, anther; Ca, carpel; DA, defective anther; DC, defective carpel. (c) Representative male and female inflorescence architecture in the same F₁. (d) Numbers of flowers per inflorescence in 50 male and 91 female individuals from the KK population. (e–j) Representative architecture of male and female inflorescences in, D. kaki (e,f), D. dygina (g,h) and D. discolor (i,j). (k–l) Similar sexual dimorphism in inflorescence architecture in Euclea pseudoebenus. Female inflorescences (k) have solitary flowers, while there are greater than three flowers in males (l). Fr, fruit. (Online version in colour.)

In one Diospyros species, D. kaki, individuals are sometimes monoecious. Wholly monoecious cultivars have the OGI gene that determines maleness in the persimmon (see above), but OGI expression is largely silenced by the insertion of a SINE-like transposon [19]. In monoecious individuals, inflorescence sex is determined by MeGI expression levels, regulated by epigenetic marks in the MeGI promoter region [19]. Importantly, monoecious D. kaki individuals bear more flowers per male inflorescence than female ones, resembling the between-individual dimorphism in persimmon (figure 3e,f) [19]. Probably, therefore, this difference between male and female inflorescences predates the evolution of dioecy.

Again using the respective native promoters, we transformed N. tabacum with MeGI (construct pMeGI-MeGI) and made lines that also had the Diospyros SuF factor, OGI gene (pMeGI-MeGI + pOGI-OGI). Recall that, in the persimmon, MeGI is fixed in the population and is not a polymorphic sex-determining gene; the transgene is therefore the mutant (m) form present in females, not the putative ancestral allele that permits male fertility in the absence of OGI. Nicotiana transformed with MeGI alone showed the previously reported male sterile phenotype [11]. As expected, this was almost completely restored by OGI (figure 4a–c; electronic supplementary material, table S3). Transformation with MeGI also induced substantially reduced flower numbers per inflorescence (to numbers like those of female persimmon plants), and introducing OGI reversed this too, increasing flower numbers to values comparable with those in the control plants (figure 4d–f). Flowering time was also affected, with MeGI producing a longer vegetative phase, while OGI reversed this (figure 4g; electronic supplementary material, table S3). OGI therefore leads to two sexually dimorphic traits having the state found in persimmon males, increased male flower numbers per inflorescence and earlier male flowering. As mentioned below, both these traits are likely to be beneficial for male functions in nature.

Figure 4.

Ectopic effects of the OGI/MeGI system in Nicotiana tabacum. (a–c) Phenotypic changes in the control transformed with the empty vector (a), the pMeGI-MeGI lines (b) and the pMeGI-MeGI + pOGI-OGI lines (c). An, anther; DA, defective anther; Pe, petal; Sg, stigma. The pMeGI-MeGI lines showed male-sterility, and the introduction of OGI could restore the sterility. (d,e) Changes in the numbers of flowers per inflorescence. (f,g) Reduced numbers of flowers per inflorescence in pMeGI-MeGI transformants (p < 0.01) (f) and delay in flowering (p < 0.01) (g), which were almost restored by introduction of the OGI. (h–l) Other phenotypic changes in the pMeGI-MeGI lines, that differ from the sexual dimorphisms observed in Diospyros. Specifically, the pMeGI-MeGI lines often bear merged flowers (MF), with two carpels but one set of anthers (n = 5) (h–l), and were occasionally dwarfed with shorter internodes (j) than in control (k); some had severe leaf serration (l), which is consistent with the phenotypes in Arabidopsis transformed with MeGI under the control of CaMV35S promoter [11]. (Online version in colour.)

We next briefly mention the third question posed above, whether M → m mutations are free from deleterious pleiotropic effects. MeGI's pleiotropic effect of reducing flower numbers need not have been deleterious. As explained in the Introduction, inbreeding depression, plus size-number trade-offs benefitting seeds, can promote the mutation's spread in a partially selfing functionally hermaphroditic population. The MeGI mutation's effect on flower number is likely to reflect a general difference in female versus male inflorescence development common to all angiosperms, not a change associated with the evolution of dioecy within the genus Diospyros [25]. A difference in inflorescence architecture is sometimes caused by members of the gene family to which MeGI belongs. This is consistent with the evidence from the breakdown species D. kaki discussed above, and data from other plants. VRS1, the MeGI orthologue of barley (Hordeum vulgare, a monocotyledonous plant distantly related to the Ebenaceae) also affects flower number per inflorescence, i.e. row number in spikelets [26]. Individuals with intact, wild-type VRS1 have solitary flowers, and are called ‘two-rowed barley’, but disruption of VRS1 (by the vrs1 mutation) results in trifurcated flowers, producing ‘six-rowed barley’. The similar phenotypic effects on inflorescence architecture in VRS1/vrs1 barley [26] and between female/male persimmon (figure 3a–c) suggest that MeGI in dioecious persimmon is a mutation in a gene that originally repressed branching in female inflorescences. Disruptions of MeGI orthologues in other angiosperms, including HB21/40/53 in Arabidopsis thaliana [27] and grassy tillers 1 (gt1) in maize [28], also promote axillary bud development or repress apical dominance, which can increase flower numbers.

As well as the effects on flower numbers and flowering time, N. tabacum transformed with MeGI often showed merged flowers (figure 4h,i, possibly related to the reduced flower number per inflorescence), shorter internodes (figure 4j,k) and leaf serration (figure 4l). In the segregating D. lotus populations, these are not consistent differences between the sexes, and therefore cannot be direct pleiotropic effects of the sex-determining genes. The changes in these traits in the transgenic plants could be related to the constitutive expression of the transgene. Importantly, changes that correspond with sexual dimorphisms seen in the dioecious species are unlikely to reflect such effects. We therefore infer that, as in Actinidia, some sexual dimorphisms in persimmon (at least the higher flower numbers and earlier flowering time in males), are pleiotropic effects of the SuF mutation (OGI in persimmon).

(c). Evolution of sexual dimorphism and recombination suppression between sex-determining regions

Our experiments also illuminate the interesting question of the evolution of sexual dimorphisms, or secondary sexual differences. Dioecious plant species often show such dimorphisms, especially in reproductive traits, including flower and inflorescence characteristics. Larger flower sizes and larger numbers per plant/inflorescence are common in males of dioecious plants, and can be advantageous for pollen dispersal in animal- and wind-pollinated plants [29–31]. In long-lived dioecious plants, including the species studied here, males commonly flower at younger ages than females [32], which could ensure pollen dispersal over female flowering times [30]. Knowing the identity of the sex-determining genes permits tests of whether the sex-determining genes themselves also control other sex differences. The assumed direct male function advantage for female-suppressing SuF factors in the two-gene models outlined above is one example, and our results show that some sexual dimorphisms indeed seem to be downstream effects of this mutation. The phenotypic differences between the sexes are shared by the two distantly related angiosperms studied, even though the upstream OGI/MeGI genes and Shy Girl acquired their sex-determining functions much more recently in the Ebenaceae and Actinidiaceae lineages [11,14]. They also resemble male/female inflorescence differences seen in monoecious species. Probably, therefore the sex differences reflect effects already present in an older developmental system that determines inflorescence sex.

If pleiotropy is important, it may explain why the persimmon and kiwifruit sex-linked regions have remained small, as sexual dimorphisms that are pleiotropic effects of the sex-determining genes themselves do not promote the evolution of such regions. By contrast, a sexually antagonistic mutation in a gene separate from the sex-determining locus may generate selection for closer linkage between the loci. After dioecy has evolved, adaptive changes may occur in genes other than the sex-determining factors, as each sex evolves in response to the new situation of unisexuality, free from constraints imposed by the other sex functions. Mutations in genes linked to the sex-determining factors are most likely to establish polymorphisms that can lead to the evolution of physically large multi-gene fully sex-linked regions [31,33–35]. However, such sex-linked regions need not evolve. Even if a mutation is sexually antagonistic, the conditions for a polymorphism to be maintained are stringent, and may often not be satisfied. For example, a male-benefit mutation may be unable to spread, either because it too strongly reduces fitness in females, or is too loosely linked to the sex-determining locus for the advantage to males to outweigh the female disadvantage. Alternatively, the advantage in one sex may be so large that the mutation spreads throughout the population and becomes fixed. This may lead to the population expressing a male-benefit trait, despite fitness-reducing effect(s) in females. Such conflicts might be resolved over evolutionary time, by changes to reduce expression of each such disadvantageous trait in the sex expressing it, eventually allowing each sex to reach its optimal phenotype.

Importantly, however, the approach used here, experimental transgenic testing of the effects of sex-determining factors in a species that is not dioecious, and has not undergone such changes, allows us to infer that the sex-determining factors themselves may sometimes produce sexual dimorphism as direct pleiotropic effects, without such subsequent evolutionary changes being involved. The most important question above (question (i)) is whether female suppressors from the two dioecious plants studied, kiwifruit and persimmon, with different molecular actions, show the expected pleiotropic effect of improved male functions. In both cases, traits that are characteristic of males in both the dioecious taxa were indeed observed in the transgenic plants. The fitness effects cannot be assessed in Nicotiana, a plant whose ecology and mating system differ from those of the species whose maleness factors we are testing. It is, however, striking that the female-suppressor mutations possessed by both Actinidia and Diospyros males have the pleiotropic effects that are assumed in theoretical models for the evolution of dioecy.

Data accessibility

This article has no additional data.

Authors' contributions

T.A. and D.C. designed the study. T.A. conducted the experiments. T.A. and D.C. analysed/interpreted the data. T.A. and D.C. drafted the manuscript and approved the manuscript.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by PRESTO (to T.A.) from Japan Science and Technology Agency (JST), and Grant-in-Aid for Scientific Research on Innovative Areas No. J16H06471 (to T.A.) from JSPS.