Sexual Dimorphism in White Campion: Complex Control of Carpel Number Is Revealed by Y Chromosome Deletions

Corresponding author: Ioan Negrutiu, ENS de Lyon, RDP - UMR 9938 CNRS/INRA/ENS, 46 Allée d'Italie, Lyon, France., ioan.negrutiu{at}ens-lyon.fr (E-mail)

Communicating editor: W. F. SHERIDAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Sexual dimorphism in the dioecious plant white campion (Silene latifolia = Melandrium album) is under the control of two main regions on the Y chromosome. One such region, encoding the gynoecium-suppressing function (GSF), is responsible for the arrest of carpel initiation in male flowers. To generate chromosomal deletions, we used pollen irradiation in male plants to produce hermaphroditic mutants (bsx mutants) in which carpel development was restored. The mutants resulted from alterations in at least two GSF chromosomal regions, one autosomal and one located on the distal half of the (p)-arm of the Y chromosome. The two mutations affected carpel development independently, each mutation showing incomplete penetrance and variegation, albeit at significantly different levels. During successive meiotic generations, a progressive increase in penetrance and a reduction in variegation levels were observed and quantified at the level of the Y-linked GSF (GSF-Y). Possible mechanisms are proposed to explain the behavior of the bsx mutations: epigenetic regulation or/and second-site mutation of modifier genes. In addition, studies on the inheritance of the hermaphroditic trait showed that, unlike wild-type Y chromosomes, deleted Y chromosomes can be transmitted through both the male and the female lines. Altogether, these findings bring experimental support, on the one hand, to the existence on the Y chromosome of genic meiotic drive function(s) and, on the other hand, to models that consider that dioecy evolved through multiple mutation events. As such, the GSF is actually a system containing more than one locus and whose primary component is located on the Y chromosome.

AMONG the species that have evolved dioecy in angiosperms, white campion is one of the rare cases in which the sexual determination (males are XY and females are XX) is both controlled by an active Y chromosome in a heteromorphic XY sex chromosome system (WESTERGAARD 1958 ; VAN NIGTEVECHT 1966 ; YE et al. 1991 ) and achieved by alterations of reproductive organ formation at early stages of development (GRANT et al. 1994 ; FARBOS et al. 1997 ). This results through the action of "female-suppressing" and "male-promoting" genes (WESTERGAARD 1958 ). Much of the available information about sexual dimorphism in this species comes from morphological, genetic, and cytogenetic studies, which strongly suggest that the male flower corresponds to a suppressed hermaphroditic condition.

Morphological and cytological studies of sexual dimorphism have precisely defined the histological changes underlying this process (FARBOS et al. 1997 ). We have shown that in the absence of the "stamen-promoting function" (SPF) putatively located on the Y chromosome, the male sexual organs undergo developmental arrest in the female flower (XX configuration). This arrest results from the lack of parietal cell initials and the degeneration of sporogenous cell initials during early anther differentiation (flower stage 6; FARBOS et al. 1997 ). In the presence of the "gynoecium-suppressing function" (GSF) located on the Y chromosome, the female sexual organs undergo developmental arrest in the male flower (XY configuration). This block results from an arrest in carpel initiation (flower stage 5; FARBOS et al. 1997 ) leading to production of a "rod-like-structure" in the center of the male flower where five carpels normally appear in the female flower. Thus, the arrest of carpel development in the male flower is the earliest event in the establishment of the sexual dimorphism in white campion (Figure 1).

View larger version (65K): In this window In a new window Download PPT slide   Figure 1. Male flowers of white campion as developmental mutants: schematic representation of GSF expression, the earliest event leading to sexual dimorphism. Alterations of GSF can be generated by pollen irradiation and result in the restoration of the ancestral hermaphroditic trait.

Establishing the linkage relationship between GSF and SPF has long been the goal of genetic and cytogenetic approaches (WESTERGAARD 1946 , WESTERGAARD 1958 ; VAN NIGTEVECHT 1966 ). From the analysis of hermaphroditic and male sterile mutants produced in crosses between diploid and polyploid lines (WESTERGAARD 1946 ) or in Silene latifolia x S. dioica interspecific crosses (VAN NIGTEVECHT 1966 ), these authors proposed a model in which GSF had a terminal location on the Y differential arm, while SPF was positioned by default on either side of the centromere. In recent work on similar mutants generated by X-ray mutagenesis, which were analyzed with physical markers hybridizing to the Y chromosome, it was proposed that "a carpel suppresser gene(s) and a stamen-promoting (i.e., male sex-determining) gene(s) must be (relatively tightly) linked on one arm of the Y chromosome" (DONNISON et al. 1996 ; LEBEL-HARDENACK and GRANT 1997 ). Despite such recent advances, we believe that the story is not resolved, mainly because previously reported mutants have not been simultaneously analyzed with respect to genetic, cytogenetic, and morphological characteristics. A further drawback is the lack of mutants affected in the SPF, whose location on the Y chromosome remains so far hypothetical (WESTERGAARD 1958 ).

We therefore decided to generate Y deletion mutants in a diploid genetic background by -irradiation of pollen. By combining irradiation of pollen (containing X or Y gametes) with screening in the M₁ generation, the Y chromosome becomes the main target in this type of mutagenesis experiment due to its permanent haploid condition. Subsequent deletion mapping could then allow the limits of the position of GSF and SPF chromosomal regions to be determined and provide appropriate tools for molecular studies. The experiments yielded asexual and hermaphroditic mutants. The former, named asx, most likely correspond to alterations in the SPF and are described in the accompanying article (FARBOS et al. 1999 , this issue). The hermaphroditic mutants should result from alterations in GSF.

In this article we report the genetic, cytogenetic, and morphological characterization of 15 independent hermaphroditic mutants, named bisexua (bsx). Their comparative analysis revealed that (1) bsx mutants were produced in a diploid genetic background at rates as high as 1%; (2) the mutants resulted from alterations in at least two GSF chromosomal regions, one autosomal (GSF-A) and one located on the Y chromosome (GSF-Y); (3) the two chromosomal mutations affected carpel development independently, each mutation showing incomplete penetrance and variegation, albeit at significantly different levels; (4) during successive meiotic generations, a progressive increase in penetrance and a reduction in variegation levels were observed and quantified; (5) the penetrance of the GSF-Y mutations was independent of the size of the corresponding Y deletions; (6) the GSF chromosomal region appeared to contain the only sporophytic reproductive function present within the distal half of the Y(p)-arm. We propose that the GSF is located near the tip of the Y(p)-arm. Possible mechanisms to explain the behavior of the bsx mutations are discussed. These findings bring experimental support to the model, which indicates that dioecy evolved through multiple mutation events.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant material, chromosomal constitution, and flower morphology:
The experiments were performed with pollen from male RAF and female MR₁ genotypes, with the exception of mutants bsx3 (male M5045 genotype). The male meiotic analysis indicated that 2n in wild type was either 24,XY or 26,XY. Most of the mutants had 2n = 24,XY. Whenever present, the extra pair of chromosomes behaved as bivalents showing no pairing with other chromosomes of the complement (see RESULTS). Similarly, the segregating population of M₁ plants produced, besides the isolated sexual mutants, only male and female segregants (Table 1). This indicated that the extra pair of chromosomes had no effect on sex expression.

The mutant hermaphroditic plants, unless otherwise stated, have bisexual flowers producing seeds following spontaneous or forced self-pollination.

Mutation induction by means of ⁶⁰Co irradiation of pollen and screening in the M₁ generation:
Pollen was irradiated at 5–50 Krad (50–500 Gy) and used directly for pollination. Germinating seeds were obtained within the 5–20 Krad range. LD50 was estimated at 11 Krad on the basis of seed germination and growth evaluations. The described mutants have been obtained at doses ranging from 5 to 11 Krad.

The screening was performed in the M₁ generation to favor the identification of mutations specifically affecting the Y chromosome. Upon irradiation, recessive mutations or deletions would only be visible if Y-linked, while mutations on X or autosomes need to be genetically dominant to be expressed in the M₁. Altered sexual phenotypes were screened visually. Additional mutant phenotypes in M₁ were occasionally identified and corresponded to albino, dwarf, or narrow-leaf plantlets.

Scanning electron microscopy (SEM):
The protocol for SEM was as described in FARBOS et al. 1997 .

Pollen viability:
Pollen was harvested at anthesis and the viability was evaluated by Alexander staining (ALEXANDER 1969 ).

Cytogenetical analysis:
For meiotic analysis, anthers of early flower buds of wild type and mutants were fixed in ethanol/acetic acid solution (3/1) and stored at -20°. The fixed material was stained and squashed as previously described (GEORGIEV 1987 ). For mitotic chromosome preparation, root tips were collected from mutant plants growing in the greenhouse and from wild-type germinated seeds and prepared according to GEORGIEV et al. 1985 .

Karyological evaluations were based on the standard karyotype as described by CIUPERCESCU et al. 1990 . The long and short arms of the Y chromosome are designated (q) and (p), respectively.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Screening of hermaphroditic mutants:
Following pollen irradiation in the M₀ generation, 2262 M₁ plants were screened for mutant sex phenotypes. Three main classes of sexual mutants were produced: 23 hermaphroditic mutants, 2 asexual, and 15 male sterile plants. The rest of the segregating plants (Table 1) were female (65%) and male (35%). The sex ratio (female/male) in the experiment was 1.86. The hermaphroditic mutant isolation rate was 1% and these mutants were named bsx (Figure 2). Nine of the bsx mutants were male and female sterile and, with one exception (bsx14), these have not been analyzed further. Genetic, cytogenetic, morphological, and functional analyses were performed on the remaining bsx mutants and are detailed below.

View larger version (134K): In this window In a new window Download PPT slide   Figure 2. Scanning electron micrographs of (A) wild-type male flower at stage 9 with tetralocular anthers and the rod-like structure surrounded by a cavity in whorl 4; (B and C) bsx-A and bsx3 mutants at the same stage 9. The first has two carpels and the second has five carpels. Certain stamens at the front have been removed in A and B to facilitate the observation of outer (antisepalloid) and inner (antipetalloid) circles of stamens. In C, the stamens start degenerating; cf. Table 3 and Table 5. Bars, 250 µm.

Genetic analysis of bsx mutants:
Selfing and/or backcrossing (BC) were performed to evaluate the capacity of the mutants to transmit the hermaphroditic trait to the progeny, but also to eliminate other potential mutations induced by the mutagenesis. The results are given in Table 1 and show that three main types of mutants were obtained: Y-linked mutations, transmitted or not transmitted to the progeny, and an autosomal mutation.

The hermaphroditic trait was transmitted to the progeny in 9 of the 14 fertile mutants analyzed. However, the hermaphroditic trait was variably inherited, from rates as low as 3% in mutants such as bsx5, bsx8, or bsx12 to rates between 20 and 50% in mutants such as bsx4, bsx10, or bsx11. Sex ratio bias, a common feature in this species (CORRENS 1928 ; TAYLOR 1994 ), bias in the transmission of a deleted Y chromosome (JANOUSEK et al. 1998 ), and deleterious mutations are most likely the main causes of these results. Eight of these 9 mutants, either when selfed or when backcrossed with the mutant as the male parent, gave only female and hermaphroditic segregants, strongly suggesting Y linkage in each case (Table 1 and Table 2). Mutant bsx-A yielded male, female, and hermaphroditic phenotypes in self- and BC progeny. We concluded that bsx-A resulted from an autosomal mutation. The mutation should therefore be dominant as it was screened in the M₁ generation. Six mutants, including bsx14, which had very low fertility (Table 1), did not transmit the trait, at least within the limits of the tested progeny populations. Because male plants did not segregate in the selfed progeny of mutants bsx6 and bsx13, or in the BC female of mutant bsx9, the results are consistent with Y linkage. The results are summarized in Table 2.

In general, the transmission rates improved in subsequent meiotic generation(s) after both selfing and backcrossing (Table 1). This indicated that the Y deletion per se was not the main reason for poor or nontransmission of the trait.

From BC evaluations of eight independent mutants we noticed that in four cases the trait was transmitted more efficiently through the female progenitor (BC male) in mutants bsx1, bsx8, bsx10, and bsx12. In two cases the male progenitor transmitted the trait more efficiently (BC female) in mutants bsx2 and bsx4. However, the low number of progeny plants tested in bsx2 reduces the significance of the corresponding results. Mutant bsx11 efficiently transmitted the trait through the male progenitor, but difficulties in emasculating the mutant did not allow testing of the transmission through the female progenitor. In mutant bsx5 the situation is so far unclear (Table 1).

In summary, among the 15 bsx mutants analyzed genetically, we were able to demonstrate the identification of 1 autosomal and 11 Y-linked mutations affecting the GSF.

Cytogenetic analysis of the bsx mutants:
To verify the diploid constitution of the mutants and to confirm the Y-linkage information generated in genetic tests, a systematic examination of meiotic configurations and of X and Y chromosomes in mitosis was undertaken. We recall that in the male meiosis there are 11 pairs of autosomes and terminally pairing (pseudoautosomal region) X and Y chromosomes. In females, 12 pairs of chromosomes are present, 1 pair of X chromosomes included. The Y chromosome is the largest chromosome of the complement (Y:X ratio = 1.4; CIUPERCESCU et al. 1990 ).

Meiotic analysis: The analysis of the male meiosis on the initial M₁ mutants is extremely important because it could provide valuable information on Y/X pairing, orientation of the Y deletions, and translocation configurations between various chromosomes. The results are presented in Table 3 and illustrated in Figure 3.

View larger version (89K): In this window In a new window Download PPT slide   Figure 3. Meiotic metaphase I spreads from (A) wild-type male plant, (B) bsx-A, (C) bsx1, (D) bsx3, (E) bsx12, (F) bsx2, (G and H) bsx5, and (I) bsx7. Arrowheads: (G) laggard chromosomes, (H) dicentric chromosome, (I) trivalents arranged in zigzag orientation. X and Y chromosomes are labeled.

In general, the meiotic chromosome pairing was normal and consisted of variable numbers of ring and rod bivalent configurations (Table 3, metaphase I column). Eleven of the 15 bsx mutants were shown to undergo normal diakinesis and metaphase I configurations, indicating that the mutagenic treatment was relatively mild. Seven mutants (bsx1, bsx4, bsx6, bsx8, bsx10, bsx13, and bsx-A) had 12 bivalents and were fertile (5 of them transmitted the trait). Thus, no aneuploidy and no large translocations or deletions had occurred in the autosomes of these mutants. Mutants bsx3 and bsx11 contained the extra pair of autosomes detected occasionally in the wild type and showed a balanced meiotic configuration. In the other mutants, various types of meiotic abnormalities were observed, of which autosomal translocations, irregular disjunction, laggards, or variable numbers of microsporocytes per tetrad were the most frequently encountered (Table 3). For example, late or occasionally dividing univalents at metaphase I or II generated laggard or dicentric chromosomes (Figure 3G and Figure H) that, at tetrad stage, were either grouped as micronuclei, as in bsx5, or formed abnormal numbers of meiocytes per tetrad, such as polyads and triads in bsx1, and pentads (Figure 3F) and hexads in bsx2 and bsx12. Autosomal associations in the form of trivalents (bsx9, bsx13, and bsx14) or quadrivalents (bsx7 and bsx14) were observed. In most cases, the trivalents were arranged in zigzag orientation at metaphase I (Figure 3I), indicating that relatively minor translocations took place. Such autosomal translocation configurations were most frequently detected in mutants bsx7, bsx9, bsx13, and bsx14. The presence of tri-, tetra-, and univalents in these mutants could be associated with male and/or female sterility, or at least with low pollen fertility (bsx9, bsx13, and bsx14; Table 3 and Table 5). In addition, important meiotic irregularities in mutants bsx7, bsx9, and bsx13 were coincident with nontransmission of the hermaphroditic trait. However, this was not the case in mutant bsx6, which had normal meiosis but did not transmit the trait. In mutants bsx3 and bsx10, the corresponding male sterility or poor pollen fertility were due to postmeiotic abnormalities.

Finally, the meiotic analysis showed that the Y chromosome was present in all the bsx mutants and that the X and Y chromosomes paired end-to-end, the conjugation occurring between the terminal homologous domains as in the wild type (Figure 3, A–E and I and Table 7). This demonstrated that the pseudoautosomal region in the homologous arm (q-arm) was intact in all the mutants.

Mitotic analysis: The X and Y chromosomes were further analyzed in mitotic preparations. The results are presented in Table 4 and illustrated in Figure 4. Arm ratio values ranged from 1.19 in the wild type to 3.31 in mutant bsx1. We demonstrate that at least 7 of the 15 analyzed mutants have measurable deletions (statistic significance level) on the Y chromosome. According to the relative size of the deletions, the mutants were classified in three groups: large deletions (>25% of the arm; bsx1 and bsx2), medium-sized deletions (between 10 and 25% of the arm; bsx 3 and bsx4), and small deletions (<10% of the arm; bsx5, bsx7, bsx9). The largest deletion affected 51% of the (p)-arm in mutant bsx1. Mutant bsx9 carries the smallest statistically confirmed deletion, i.e., 5% of the (p)-arm. In the remaining 7 bsx mutants, the altered GSF of the Y might correspond to a small or very small deletion. bsx mutants with small deletions on the Y represented the majority of the mutants obtained. In the case of the autosomal mutant bsx-A, a small deletion (Table 4) on the differential arm of the Y cannot be ruled out, but such a deletion must be silent and should not involve the Y-linked GSF, as indicated by the genetic data (Table 1 and Table 2).

View larger version (76K): In this window In a new window Download PPT slide   Figure 4. Mitotic metaphase spreads from (A) wild-type male plant, (B) bsx-A, (C) bsx1, (D) bsx7, and (E) bsx12. X and Y chromosomes are labeled.

(p)-arm location of the deletions: Because X and Y chromosomes paired terminally in meiosis in all bsx mutants (Table 3 and Table 7) and the [Y(q)-arm length/total X length] ratio did not significantly vary between wild type and mutants (not shown), we conclude that the (q)-arm was not modified in bsx mutants. The deletions have most likely affected the shorter differential (p)-arm. We evaluated the breakpoints by performing in situ hybridization on chromosomes with the subtelomeric probe X.43 (FARBOS et al. 1999 , this issue) in mutants bsx3 and the asexual mutants asx1 and asx2. These mutants share the same parental Y chromosome (cf. MATERIALS AND METHODS). The results are presented in the accompanying article (FARBOS et al. 1999 ) and show that bsx3, but not the asx mutants, has a terminal deletion on the (p)-arm.

Morphological and functional analysis of the mutants:
The following traits were considered for comparison between bsx mutants and the wild type: pollen fertility (evaluated by both Alexander staining and seed-set tests in hand pollinations), carpel number, percentage of male flowers, and seed set in self-pollination (spontaneous and forced). We first determined that tight correlations existed between style number and carpel number or pollen viability and seed set in crosses of wild type as female x mutant as male (not shown). The results are outlined in Table 5, with carpel number and pollen fertility representing the most important traits. Pollen viability varied widely in the M₁ generation, from male sterility (mutants bsx3 and bsx14) to high fertility (mutants bsx-A, bsx6, bsx8, bsx11). Nine of the 15 mutants had a pollen viability >25%.

Carpel number estimates in M₁ generation showed variation both between and within the mutants. In only one mutant (bsx3) was the average carpel number similar to the wild-type female. In the other mutants, the average carpel number per flower varied from 0.5 in bsx-A to 4.6 in mutant bsx6. In the case of bsx-A, the first flowers on the inflorescence were bisexual (Figure 2). Subsequently, male flowers were produced abundantly, the incidence of bisexual flowers rapidly decreasing; mosaic patterns consisted of isolated or very small sectors of bisexual flowers occurring in an unpredictable manner. Eight mutants had average carpel numbers equal to or above four. Usually, mutants having average carpel numbers below four also produced male flowers (andromonoecious-type plant). This ranged from 67% in mutant bsx-A to 2% in mutant bsx7. Occasional production of male flowers was observed in mutants bsx4 and bsx11, which had average carpel numbers of four or more. Similar evaluations were performed in progeny plants derived from selfing (M₂ generation) or from BC. Interestingly, with the exception of bsx-A, the average carpel number per flower increased in one generation by approximately one unit in all the mutants tested. At the same time, these mutants stopped producing male flowers. In summary, the results showed that (1) the mutation or the deletion of the GSF resulted in a defined level of carpel number restoration in each mutant in the M₁ generation (average number, 3.6; range, 1 to 5, bsx-A not included); (2) a similar variation was registered within individual mutants, the variation pattern being unpredictable; (3) passage through one meiotic cycle resulted in a systematic increase in carpel number; and (4) variation in carpel number had no effect on other flower parts.

Seed set was evaluated as a general indicator of fertility. The analysis consisted of establishing the number of seeds/capsule after selfing and BC. Only data from selfing are presented in Table 6. The main trend was irregular seed setting in spontaneous selfing in the M₁ generation. One possible explanation is the wide variation in carpel number in most M₁ mutants. Table 6 shows that there is a strong correlation between carpel number and seed set. Another observation, more difficult to quantify, was that M₁ plants frequently had tightly folded petals, which prevented pollen access to the styles. In general, seed set was significantly improved in subsequent generations.

Integrated analyses of genetic, cytogenetic, and morphological data:
The genetic, cytogenetic, and morphological data on 15 bsx mutants is summarized in this section. We classify the mutants according to the size of the Y deletions and compare them through their main reproductive features, such as meiosis (X/Y pairing and type of meiotic abnormalities), pollen viability, carpel number, and inheritance of the bsx trait (Table 7).

The results show that no correlations exist between the size of the Y deletions and pollen viability, the transmission of the hermaphroditic trait to the progeny or carpel number. For example, meiotic abnormalities did not correlate with other analyzed traits, except for pollen viability. This implies that meiotic and pollen defects, when observed, were not a direct consequence of Y deletions that impaired the GSF. Furthermore, the last column in Table 7 clearly shows that the extent of Y deletions and transmission rates are not related.

Concerning carpel development, several points are important. The overall variation in average carpel number per flower between independent mutants varies by a factor of 10: in mutant bsx-A, 0.5 carpels per flower corresponding to 10% penetrance of the GSF mutation and in mutant bsx3, 5 carpels per flower, i.e., a full penetrance of the mutation. This variation was only fivefold among the Y deletion series of bsx mutants. Incomplete penetrance of the mutation occurred frequently. However, >50% of the mutants produced an average of at least 4 carpels per flower in the M₁ generation, indicating that the deletion of the GSF chromosomal region generally resulted in a significant restoration of carpel development.

The carpel number was observed to vary within each mutant with production of male flowers (i.e., zero carpels) being registered in 8 out of 15 mutants evaluated (andromonoecious type). This was an important observation, at least in the case of mutants with GSF-Y deletions. Mutants bsx1, bsx5, bsx8, and bsx9 had a relatively low initial average carpel number (approximately three) and produced male flowers at rates varying from 3 to 12%. We concluded that the andromonoecious type essentially corresponded to mutants with weak penetrance of the mutation, in which flower meristems or small inflorescence sectors randomly inherited a wild-type level of the Y-linked GSF activity. A significantly different strength in penetrance and variegation levels was observed in the autosomal mutant bsx-A, in which three-quarters of the flowers were male, approximately sixfold more than in Y-deleted bsx mutants of the andromonoecious type. On the basis of its low average carpel number per flower and lack of clear-cut deletion on the Y chromosome, bsx14 also appeared to be a potential candidate for an autosomal GSF mutation.

The combined cytogenetic and morphological analyses of the bsx mutants generated helpful data on the GSF chromosomal regions. Further work will concentrate on mutants bsx1, bsx3, bsx4, bsx11, bsx12, and bsx-A. The expected outcome is a more precise mapping of the deletions and, implicitly, of the GSF-Y chromosomal region. It should also reveal the regulatory relationship between the autosomal (GSF-A) and the Y-linked GSF (GSF-Y).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have analyzed 15 hermaphroditic mutants in the dioecious white campion. These mutants were produced by -irradiation of pollen and their frequency was estimated at 1% of the M₁ screened population. The rate is 10-fold higher than that of any other class of sexual mutants produced in the experiment (FARBOS et al. 1999 , this issue). This could indicate a high incidence of terminal, vs. interstitial, deletions under the experimental conditions used (LEE and KAMRA 1981 ; LEFRANCOIS et al. 1989 ; VIZIR et al. 1994 ). Mutants with similar characteristics were first described by WESTERGAARD 1958 , but no such mutants so far have been generated in diploid backgrounds and analyzed in detail at the genetic, cytogenetic, and morphological level.

This approach has enabled us to distinguish two chromosomal regions with GSF properties: a Y-linked and an autosomal region. No bsx mutants in an XX background were identified. Y-linked and potentially Y-linked mutants represented by far the largest group (13 out of 15), as expected due to the experimental design. One dominant autosomal mutation was obtained at a frequency at least 10-fold lower than the Y-linked ones and is of particular importance for further studies (see below). The corresponding autosomal wild-type function must operate, like the Y-linked GSF, as a negative regulator of carpel formation. The two functions may be superimposed on the carpel developmental pathway in male plants. Alternatively, the autosomal GSF could be an intrinsic component of this pathway.

The Y-deletion series presented deletions ranging from small to as large as half the length of the Y chromosome (p)-arm. Interestingly, the largest deletions accurately described so far in hermaphroditic mutants in white campion were in the same size range (WESTERGAARD 1958 ; VAN NIGTEVECHT 1966 ). It is possible that larger deletions are very unstable or that critical function(s) are located proximal to the deletion breakpoint in mutant bsx1 (FARBOS et al. 1999 , this issue). An early SPF has long been postulated to be located on the Y chromosome (WESTERGAARD 1958 ; VAN NIGTEVECHT 1966 ). Attempts to map the SPF and GSF chromosomal regions have already been made and are discussed in FARBOS et al. 1999 (this issue). Our results allow us to go one step further. Because the size of the Y deletions did not correlate with the penetrance of GSF mutations in general and because five mutants with very small deletions on the Y chromosome (bsx6, bsx7, bsx11, bsx12, and bsx 13) show high penetrance in the M₁ generation, and bsx11 and bsx12, which transmit the trait, exhibit near complete penetrance after one or two meiotic generations, we believe that the GSF-Y represents a defined region on the (p)-arm, with no corresponding redundancy present elsewhere along the distal half of the arm. Therefore, the small deletions must have affected the same region on the (p)-arm. Because mutant bsx1, which shows the largest deletion, was male and female fertile, we deduced that the deleted arm fragment did not contain other essential gene(s) involved in sporophytic reproductive development. Finally, on the basis of comparative analysis of bsx3 and asx mutants (see FARBOS et al. 1999 , this issue), we postulate that the GSF-Y region is located toward the tip of the (p)-arm, which is in disagreement with data of DONNISON et al. 1996 .

Do our results indicate whether other genes are present on the distal (p)-arm region? Table 1 shows deviations from wild-type sex ratio bias. While sex ratio control in white campion is genetically variable, most populations (TAYLOR 1994 ), including ours, exhibit female bias. One possible reason for this bias is the inability of the wild-type Y chromosome to pass through the female line (JANOUSEK et al. 1998 ; also see LYTTLE 1993 ). Here we show that Y-deletions that generate hermaphroditic mutant phenotypes can transmit the trait through the male and/or the female parent and, at the same time, exhibit modified sex ratio bias. On the basis of these results, we hypothesize the existence of female gametophytic lethality factors on the distal part of the Y(p)-arm that could act either as enhancers of the bias (by reducing the proportion of males in the population) or simply as deleterious alleles. The fact that in interspecific crosses the sex ratio bias is increased further indicates that these factors are enhancers of bias rather than deleterious alleles (TAYLOR 1993 ). Autosomal factors may need to be taken into consideration in explaining the range of modified sex ratio bias we observed in the analyzed bsx mutants.

The reported results provide further information on the possible function and mechanism(s) of action of GSF. The male plant (XY) contains all the genetic information required to produce a bisexual flower. The GSF system exerts a particularly strong and stable negative regulatory effect on carpel initiation: male flowers contain no carpels.

Alterations in GSF chromosomal regions have generated variable states of carpel number restoration, both between and within the mutants. Such variations have been noticed by several authors (VAN NIGTEVECHT 1966 ; DONNISON et al. 1996 ; JANOUSEK et al. 1996 ), but the present study uses a carpel number bioassay to reveal the significance of the process and to increase understanding of the GSF for the first time. Three explanations have been envisaged for the inter- and intra-mutant variations in carpel number restoration.

The first has the merit of integrating the complete penetrant phenotype of bsx3 mutant and assumes that it is the only mutant in which the deletion has actually eliminated the GSF-Y locus. In the other bsx mutants, the deletions may have positioned the GSF chromosomal region in new chromatin environments, giving PEV (position-effect-variegation)-like phenotypes (KARPEN 1994 ; DERNBURG et al. 1996 ). The fact that bsx3 does not transmit the trait poses problems for the testing of this hypothesis. In addition, the fact that the penetrance of GSF mutations increases in bsx mutants in successive meiotic generations gives less support to this interpretation.

The second explanation assumes that the deletion of the GSF chromosomal region per se results in variable states of carpel number restoration, both between and within the mutants. The physical loss of the Y-linked GSF appeared insufficient to restore carpel number in the M₁ generation. Thus, additional factor(s), present somewhere else in the genome, may synergistically interact with the Y-linked GSF to achieve the complete inhibition of carpel development in wild-type male flowers. bsx-A may correspond to a partially redundant component. This is supported by the fact that the Y deletion series of bsx mutants and the autosomal bsx-A mutant independently allow carpel development when mutated, albeit at significantly different levels. Interestingly, the GSF-A only seems to be active in the presence of the Y, since no autosomal GSF is operating in XX female flowers. The analysis of double mutants combining the Y-linked and the autosomal locus should clarify this point.

How might such a GSF system work at the molecular level to explain the control of carpel number through factors acting as specific negative regulators of cell proliferation in whorl 4 (FARBOS et al. 1997 ; MEYEROWITZ 1997 )? The mechanism(s) should also explain the progressive nature of the process through successive meiotic generations, perhaps due to the reduction or elimination of the remnant effect of GSF. Two such regulatory mechanisms could account for the reported patterns of GSF(s): site-specific methylation or para-mutation. Thus, the GSF-Y could act through an epigenetic mechanism—with clonal inheritance of a particular methylation state in all sister cells (EDEN and CEDAR 1994 ; RONEMUS et al. 1996 )—on specific target genes that control carpel development. One known example is the site-specific methylation of the SUPERMAN gene in Arabidopsis, which generated increased carpel numbers in the flower (JACOBSEN and MEYEROWITZ 1997 ). Methylation has recently been proposed as a possible mechanism for GSF regulation of target genes in white campion based on 5-azaC treatments applied to germinating seedlings and resulting in mosaic-patterned hermaphroditic XY plants (JANOUSEK et al. 1996 ). The treatment most likely circumvented the GSF-Y by (re)activating possible target genes on autosomes that control carpel development.

The presence of two partially redundant GSF chromosomal regions (GSF-Y and GSF-A) makes possible an alternative (epigenetic) mechanism of GSF action, namely, para-mutation (MATZKE et al. 1996 ). Such a regulatory process, together with bona fide methylation, could be an important component of dynamic developmental regulatory systems. In such a context, the Y-linked locus could act as the main para-mutagenic locus.

The third explanation of the reported patterns of GSF(s) concerns mutations at modifier loci. The transacting modifier gene hypothesis implies that second-site mutations occur either on Y or on the autosomes (or both). Few modifier gene systems are known in plants, one of the best characterized being the one affecting the opaque2 mutation in maize (LOPES et al. 1995 ). The identified modifiers can act semidominantly and are under developmental control. Dosage dependence of modifier genes is clearly established in Drosophila (HENIKOFF 1996 ). Whether such systems are also operating in the GSF system is not yet clear, but several experimental facts can be analyzed in such a context: (1) All mutants, with the exception of bsx3, exhibit incomplete penetrance and variable expressivity, while no correlation exists between the size of the Y-deletions and the observed variability. As a consequence, modifier genes need to be either relatively tightly linked to GSF or represent autosomal genes systematically associated with GSF mutations. An alternative explanation could be that altered modifiers were already present in the genetic background of some of our lines. Interestingly, bsx3 belongs to a different genetic background than the other lines. (2) The systematic increase in carpel number in successive sexual generations—both in self and in BC, which indicates no dosage effect—strongly suggests an epigenetic component and, as shown in Arabidopsis, genetic modifiers that impair chalcone synthase silencing operate by reducing DNA methylation (FURNER et al. 1998 ). The analysis of numerous hermaphroditic segregants in the M₂ generation of mutants such as bsx1 and bsx4 should allow the identification of putative modifier factors.

Taken together, the GSF system in white campion, as described herein, supports the "incomplete dominance" hypothesis, according to which full dioecy was established by more than one mutation in one sexual pathway (CHARLESWORTH 1996 ). Here we show that independent mutations at two GSF chromosomal regions give dominant restoration of carpel development. The Y-linked GSF provides the dominant female sterile mutation f^S predicted by models of the evolution of separate sexes and generation of "primitive" sex chromosomes (Y/X is therefore f^Sm^F/f^fm^s; CHARLESWORTH 1996 ). The reported results do not support the existence of a corresponding recessive GSF on the X, the Y/X formula actually being f^Sm^F/-m^s. Instead, an autosomal GSF operates interactively with the Y locus and becomes an obligate element in any models attempting to explain GSF activity and evolution. The existence of putative modifier loci adds an additional dimension to the GSF system.

In conclusion, the detailed analysis of bsx mutants produced in a diploid background has enabled us to refine, at genetic, cytogenetic, and morphological levels, the present understanding of GSF in white campion. In addition, we put forward plausible molecular mechanisms through which a negative regulation of carpel initiation has evolved, most likely in more than one step, to establish one key component—the GSF system—of sexual dimorphism in white campion.

	ACKNOWLEDGMENTS

The authors are indebted to Richard Blanc and Hervé Leyral for excellent technical assistance. They are much indebted to Deborah Charlesworth, Sheila McCormick, Gwyneth Heckel, Françoise Monéger, and Charlie Scutt for helpful discussions and/or critical reading of the manuscript. G.S. is a fellow of Centre International des Etudiants et Stagiaires and Professeur Associé Temporaire research programs of the Ministry of Foreign Affairs and Ministry of Education.

Manuscript received May 11, 1998; Accepted for publication December 8, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALEXANDER, M. P., 1969  Differential staining of aborted and non-aborted pollen. Stain Technol. 44:117-122[Medline].

CHARLESWORTH, B., 1996  The evolution of chromosomal sex determination and dosage compensation. Curr. Biol. 6:149-162[Medline].

CIUPERCESCU, D. D., J. VEUSKENS, A. MOURAS, D. YE, and M. BRIQUET et al., 1990  Karyotyping Melandrium album, a dioecious plant with heteromorphic sex chromosomes. Genome 33:556-562.

CORRENS, C., 1928 Bestimmung, vererbung und verteilung des geschlechtes bei höheren planzen, pp. 1–138 in Hanbd. Vererbungswissensch. Edited by C. BORNTRAEGER. Verlag, Berlin.

DERNBURG, A. F., K. W. BROWAN, J. C. FUNG, W. F. MARSHALL, and J. PHILIPS et al., 1996  Perturbation of nuclear architecture by long-distance chromosome interactions. Cell 85:745-759[Medline].

DONNISON, I. S., J. SIROKY, B. VYSKOT, H. SAEDLER, and S. R. GRANT, 1996  Isolation of Y chromosome-specific sequences from Silene latifolia and mapping of male sex-determining genes using representational difference analysis. Genetics 144:1893-1901[Abstract].

EDEN, S. and H. CEDAR, 1994  Role of DNA methylation in the regulation of transcription. Curr. Opin. Genet. Dev. 4:255-259[Medline].

FARBOS, I., M. OLIVEIRA, I. NEGRUTIU, and A. MOURAS, 1997  Sex organ determination and differentiation in the dioecious plant Melandrium album (Silene latifolia): a cytological and histological analysis. Sex. Plant Reprod. 10:155-167.

FARBOS, I., J. VEUSKENS, M. OLIVEIRA, B. VYSKOT, and S. HINNISDAELS et al., 1999  Sexual dimorphism in white campion: deletion on the Y chromosome results in a floral asexual phenotype. Genetics 151:1187-1196[Abstract/Free Full Text].

FURNER, I. J., M. A. SHEIKH, and C. E. COLLETT, 1998  Gene silencing and homology-dependent gene silencing in Arabidopsis: genetic modifiers and DNA methylation. Genetics 149:651-662[Abstract/Free Full Text].

GEORGIEV, S., 1987  Cytogenetic studies of the nature of the sphaerococcum mutation in genus Triticum L. Genet. Breed. 19:95-105.

GEORGIEV, S., K. GECHEFF, G. KÜNZEL, and R. RIEGER, 1985  Giemsa N-banding as a tool for identification of chromosomes reconstruction in barley. Biologische Zentr. Blat. Berlin 104:29-34.

GRANT, S., B. HUNKIRCHEN, and H. SAEDLER, 1994  Developmental differences between male and female flowers in the dioecious plant Silene latifolia.. Plant J. 6:471-480.

HENIKOFF, S., 1996  Dosage-dependent modification of position- effect variagation in Drosophila. BioEssay 18:401-409[Medline].

JACOBSEN, S. E. and E. M. MEYEROWITZ, 1997  Hypermethylated SUPERMAN epigenetic alleles in Arabidopsis.. Science 277:1100-1103[Abstract/Free Full Text].

JANOUSEK, B., J. SIROKY, and B. VYSKOT, 1996  Epigenetic control of sexual phenotype in a dioecious plant, Melandrium album.. Mol. Gen. Genet. 250:483-490[Medline].

JANOUSEK, B., S. GRANT, and B. VYSKOT, 1998  Non-transmissibility of the Y chromosome through the female line in androhermaphrodite plants of Melandrium album.. Heredity 80:576-583.

KARPEN, G. H., 1994  Position-effect-variegation and the new biology of heterochromatin. Curr. Opin. Genet. Dev. 4:281-291[Medline].

LEBEL-HARDENACK, S. and S. R. GRANT, 1997  Genetics of sex determination in flowering plants. Trends Plant Sci. 2:130-136.

LEE, C. L. and O. P. KAMRA, 1981  The pattern of radiation-induced transmissible aberrations in a human cell culture. Hum. Genet. 57:380-384[Medline].

LEFRANÇOIS, D., W. AL ACHKAR, A. AURIAS, J. COUTURIER, and A. M. DUTRILLAUX et al., 1989  Chromosomal aberrations induced by low-dose -irradiation: study of R-banded chromosomes of human lymphocytes. Mutat. Res. 212:167-172[Medline].

LOPES, M. A., K. TAKASAKI, D. E. BOSTWICK, T. HELENTJARIS, and B. A. LARKINS, 1995  Identification of two opaque2 modifier loci in quality protein maize. Mol. Gen. Genet. 247:603-613[Medline].

LYTTLE, T. W., 1993  Cheeters sometimes prosper: distortion of Mendelian segregation by meiotic drive. Trends Genet. 9:205-210[Medline].

MATZKE, M. A., A. J. M. MATZKE, and W. B. EGGLESTON, 1996  Paramutation and transgene silencing: a common response to invasive DNA? Trends Plant Sci. 1:382-388.

MEYEROWITZ, E. M., 1997  Genetic control of cell division patterns in developing plants. Cell 88:299-308[Medline].

RONEMUS, M. J., M. GALBIATI, C. TICKNOR, J. C. CHEN, and S. L. DELLAPORTA, 1996  Demethylation-induced developmental pleiotropy in Arabidopsis.. Science 273:654-657[Abstract].

TAYLOR, D. R., 1993  Sex ratio in hybrids between Silene alba and Silene dioica: evidence for Y-linked restorers. Heredity 74:518-526.

TAYLOR, D. R., 1994  The genetic basis of sex ratio in Silene alba (= S. latifolia). Genetics 136:641-651[Abstract].

VAN NIGTEVECHT, G., 1966  Genetic studies in dioecious Melandrium: I. Sex-linked and sex-influenced inheritance in Melandrium album and Melandrium dioicum. Genetica 37:281-306.

VIZIR, I. Y., M. L. ANDERSON, Z. A. WILSON, and B. J. MULLIGAN, 1994  Isolation of deficiencies in the Arabidopsis genome by -irradiation of pollen. Genetics 137:1111-1119[Abstract].

WESTERGAARD, M., 1946  Aberrant Y chromosomes and sex expression in Melandrium album.. Hereditas 32:419-443.

WESTERGAARD, M., 1958  The mechanism of sex determination in dioecious flowering plants. Adv. Genet. 9:217-281[Medline].

YE, D., M. OLIVEIRA, J. VEUSKENS, Y. WU, and P. INSTALLÉ et al., 1991  Sex determination in the dioecious Melandrium: the X/Y chromosome system allows complementary cloning strategy. Plant Sci. 80:93-106.

This article has been cited by other articles:

				R. Bergero, D. Charlesworth, D. A. Filatov, and R. C. Moore Defining Regions and Rearrangements of the Silene latifolia Y Chromosome Genetics, April 1, 2008; 178(4): 2045 - 2053. [Abstract] [Full Text] [PDF]

				A. Koizumi, Y. Amanai, K. Ishii, K. Nishihara, Y. Kazama, W. Uchida, and S. Kawano Floral Development of an Asexual and Female-Like Mutant Carrying Two Deletions in Gynoecium-Suppressing and Stamen-Promoting Functional Regions on the Y Chromosome of the Dioecious Plant Silene latifolia Plant Cell Physiol., October 1, 2007; 48(10): 1450 - 1461. [Abstract] [Full Text] [PDF]

				J. Zluvova, S. Georgiev, B. Janousek, D. Charlesworth, B. Vyskot, and I. Negrutiu Early Events in the Evolution of the Silene latifolia Y Chromosome: Male Specialization and Recombination Arrest Genetics, September 1, 2007; 177(1): 375 - 386. [Abstract] [Full Text] [PDF]

				J. Zluvova, M. Nicolas, A. Berger, I. Negrutiu, and F. Moneger Premature arrest of the male flower meristem precedes sexual dimorphism in the dioecious plant Silene latifolia PNAS, December 5, 2006; 103(49): 18854 - 18859. [Abstract] [Full Text] [PDF]

				M. Tanurdzic and J. A. Banks Sex-Determining Mechanisms in Land Plants PLANT CELL, June 1, 2004; 16(suppl_1): S61 - S71. [Full Text] [PDF]

				M. Lengerova, R. C. Moore, S. R. Grant, and B. Vyskot The Sex Chromosomes of Silene latifolia Revisited and Revised Genetics, October 1, 2003; 165(2): 935 - 938. [Abstract] [Full Text] [PDF]

				R. C. Moore, O. Kozyreva, S. Lebel-Hardenack, J. Siroky, R. Hobza, B. Vyskot, and S. R. Grant Genetic and Functional Analysis of DD44, a Sex-Linked Gene From the Dioecious Plant Silene latifolia, Provides Clues to Early Events in Sex Chromosome Evolution Genetics, January 1, 2003; 163(1): 321 - 334. [Abstract] [Full Text] [PDF]

				M. L. BUIDE and J. GUITIAN Breeding System in the Dichogamous Hermaphrodite Silene acutifolia (Caryophyllaceae) Ann. Bot., December 1, 2002; 90(6): 691 - 699. [Abstract] [Full Text] [PDF]

				C. P. Scutt, T. Jenkins, M. Furuya, and P. M. Gilmartin Male Specific Genes from Dioecious White Campion Identified by Fluorescent Differential Display Plant Cell Physiol., May 15, 2002; 43(5): 563 - 572. [Abstract] [Full Text] [PDF]

				S. Lebel-Hardenack, E. Hauser, T. F. Law, J. Schmid, and S. R. Grant Mapping of Sex Determination Loci on the White Campion (Silene latifolia) Y Chromosome Using Amplified Fragment Length Polymorphism Genetics, February 1, 2002; 160(2): 717 - 725. [Abstract] [Full Text] [PDF]

				I. Atanassov, C. Delichere, D. A. Filatov, D. Charlesworth, I. Negrutiu, and F. Moneger Analysis and Evolution of Two Functional Y-Linked Loci in a Plant Sex Chromosome System Mol. Biol. Evol., December 1, 2001; 18(12): 2162 - 2168. [Abstract] [Full Text] [PDF]

				I. Negrutiu, B. Vyskot, N. Barbacar, S. Georgiev, and F. Moneger Dioecious Plants. A Key to the Early Events of Sex Chromosome Evolution Plant Physiology, December 1, 2001; 127(4): 1418 - 1424. [Full Text] [PDF]

				D. E. Wolf, J. A. Satkoski, K. White, and L. H. Rieseberg Sex Determination in the Androdioecious Plant Datisca glomerata and Its Dioecious Sister Species D. cannabina Genetics, November 1, 2001; 159(3): 1243 - 1257. [Abstract] [Full Text] [PDF]