Molecular genetics, physiology and biology of self-incompatibility in Brassicaceae

Abstract

Self-incompatibility (SI) is defined as the inability to produce zygotes after self-pollination in a fertile hermaphrodite plant, which has stamens and pistils in the same flower. This structural organization of the hermaphrodite flower increases the risk of self-pollination, leading to low genetic diversity. To avoid this problem plants have established several pollination systems, among which the most elegant system is surely SI. The SI trait can be observed in Brassica crops, including cabbage, broccoli, turnip and radish. To produce hybrid seed of these crops efficiently, the SI trait has been employed in an agricultural context. From another point of view, the recognition reaction of SI during pollen-stigma interaction is an excellent model system for cell-cell communication and signal transduction in higher plants. In this review, we describe the molecular mechanisms of SI in Brassicaceae, which have been dissected by genetic, physiological, and biological approaches, and we discuss the future prospects in relation to associated scientific fields and new technologies.

1. Introduction

In most species of plants the flowers are hermaphrodite that is both stamen and pistil are present in the same flower. Self-pollination occurs easily in hermaphrodite flowers, and leads to inbreeding depression and decreased genetic variation; therefore plants have evolved several mechanisms to avoid self-pollination. Self-incompatibility (SI) is defined as “the inability of a fertile hermaphrodite seed plant to produce zygotes after self-pollination”, which is surely the most elegant pollination system.¹⁾ This trait attracted the attention of Charles Darwin, who published detailed descriptions of pollination and the form of flowers during his studies of evolution.^2,3) In the past, the SI phenotype was mostly evaluated by seed formation after open pollination (artificial self-pollination and cross-pollination). However, from the evaluation of seed formation, the phenotype of “incompatibility” and “sterility” cannot be distinguished. However, in relation to biological events, “incompatibility” and “sterility” are clearly different. “Sterility” is caused by non-functional male and/or female components, whereas in “incompatibility” there is a lack of seed formation in a specific male and female combination, both of which are functional. Because the final phenotypes in “incompatibility” and “sterility” are quite similar, this phenomenon was originally termed “self-sterility”. Taking the initial letter of “sterility”, the genetic locus regulating SI was termed the “S locus”, which is still used as the authorized locus name.

SI is classified morphologically into heteromorphic SI and homomorphic SI. Heteromorphic SI is related to different flower shapes; for example, pin (long pistil and short stamen length) and thrum (short pistil and long stamen length) forms are present in primrose (Primula species), buckwheat (Fagophyrum species) and star fruit (Averrhoa species). Cross pollination between different flower types is compatible and is regulated by a single locus with two alleles, S and s (Fig. 1). The genotype of pin is ss homozygote, and that of thrum is Ss heterozygote. Thus, on cross-pollination, a frequency of 50% each of pin and thrum should be maintained in a wild population under SI regulation. Interestingly, this S gene controls both the incompatibility phenotype and the flower shape (that is length of pistil and stamens). Although several attempts have been made to dissect genes residing at the S locus in plants with heteromorphic SI, no S genes or their products have been identified to date.¹⁾

Figure 1.

Photograph (A) and schematic model (B) of heteromorphic SI in primrose. (A) The pin (long pistil and short stamen length) form is shown on the left, and thrum (short pistil and long stamen length) is on the right. (B) Selfed pollen and pollen derived from the same flower form are incompatible. Crossed pollen and pollen derived from the different flower form are compatible.

On the other hand, homomorphic SI species have a single flower shape and are classified by the inheritance pattern of the S gene into gametophytic SI and sporophytic SI.^1,4) Both SI phenomena are controlled by a single locus, with multiple S alleles (S¹, S², S³, … Sⁿ) in most of the plant species. As described in Fig. 2, the incompatible phenotype is apparent in self-pollination and in cross-pollination between pollen and stigma of two different plants carrying the same S allele. To represent alleles, superscripts are commonly used in general genetic research; however, traditionally, S alleles are denoted by subscripts. In most of our papers and review articles, we have used superscript allele symbols, according to the general genetic convention.^4–8) Therefore, in this article, we have again used superscripts for allelic gene representation.

Figure 2.

Schematic model of the S locus. The S locus contains at least two genes, one encoding the male S determinant, which is expressed in anther tapetum surrounding the pollen grains during development, and the other encoding the female S determinant, which is expressed in the pistil (stigma). Both the genes encoding male and female S determinants are inherited as one segregating unit. The number of S alleles is thought to be over 50 to 100 in Brassica species.^167–169) When both male and female S determinants are derived from the same S alleles, the incompatible response occurs. In contrast, in the case of a combination of different S alleles from the male and female parents, the compatible response is observed.

In gametophytic SI (GSI), the S phenotype of pollen is determined by its own haploid S gene. The self-pollen inhibition occurs during pollen tube elongation in the style of the pistil in most GSI. Plant species in families Fabaceae, Onagraceae, Poaceae, Rosaceae, Plantaginaceae and Solanaceae exhibit GSI (Fig. 3A). On the other hand, in sporophytic SI (SSI), the S phenotype of pollen is determined by its parental diploid S gene interaction, so that dominance relationships occur between S alleles (Fig. 3B).^9,10) The landmark genetic SSI model, established by Bateman from a series of experiments, is an excellent explanation of S gene behavior, and is still valid today.^11–13) The molecular mechanisms of S allele dominance relationships will be described below. In most SSI plant species, from families Asteraceae, Brassicaceae and Convolvulaceae, pollen tube inhibition occurs on the stigma surface (Fig. 4).

Figure 3.

Relationships between phenotype and genotype in GSI (A) and SSI (B). The pistil (stigma) phenotype presented here is S¹S³ codominance, to clearly demonstrate the difference between the GSI and SSI. In both types of SI, pollen is rejected when the SI phenotype of pollen is identical to that of the pistil (stigma). (A) In the case of GSI, the pollen phenotype is identical to the genotype. The pistil phenotype of the GSI is always expressed in codominance. (B) In the case of SSI, the pollen phenotype is determined by interaction of the parental genotype; therefore, there is a dominance relationship at the pollen side. If S¹ is dominant over S², all pollen grains produced from the S¹S² pollen parent show S¹ phenotype, and are rejected on the S¹S³ (S¹ = S³) stigma. In contrast, on the same stigma (S¹ = S³), if S³ is recessive to S⁴ at the pollen side, all pollen grains, which show S⁴ phenotype, germinate and pollen tube penetration is observed.

Figure 4.

Micrographs showing (A) the inhibition of self-pollen and the inability of the emerging pollen tubes to invade the papilla cell wall (incompatibility), and (B) the penetration of cross-pollen tubes into the papilla cells (compatibility) in Brassica rapa L. Yellow arrow indicates the penetrated pollen tubes. With aniline blue staining, pollen tubes appear as distinct tubes with prominent dots, because callose plugs in the pollen tubes fluoresce brightly. In contrast, vascular bundles and other tissues do not fluoresce brightly, as shown in (A). Bar = 200 µm. (Courtesy of Masaaki Osaka and Ken-ichiro Hiroi, unpublished data).

SI is interesting in two respects: firstly, SI involves cell-cell communication between pollen/pollen tube and pistil, and signal transduction in the stigma/style; and, secondly, SI is an important trait for F₁ hybrid seed production in Brassica rapa (syn. campestris, e.g. Chinese cabbage and turnip), and B. oleracea (e.g. cabbage, cauliflower, broccoli, kohlrabi, kale and Brussels sprouts), which was established in a Japanese seed company before the 2nd World War.⁶⁾ In other words, we are interested in Brassica SI from the viewpoint of both fundamental biology and agricultural applications. Thus, an understand of the molecular mechanisms of SI in Brassica species is an important research target. In this review article, we will summarize SI research in Brassicaceae species, including molecular cloning of the male and female S determinants, epigenetic regulation of dominance relationships between S alleles, and evolution of the SI system. Finally, we will discuss the future prospect for SI research. Many review articles have been published, and are helpful in understanding this field.^{1,4–8,14,15)}

2. Identification of the S-locus-specific proteins from stigma

In the early SI research, Oenothera (Onagraceae), Malus, Prunus (Rosaceae), Nicotiana, Solanum, Petunia (Solanaceae), Trifolium (Leguminosae), Secale, Pahalaris, Fastuca (Poaceae), Cardamine, Iberis, Brassica (Brassicaceae), Crepis, Cosmos (Asteraceae), Theobroma (Sterculiaceae), Ipomoea (Convolvulaceae) and Antirrhinum (Plantaginaceae), among others, were studied.¹⁾ These species were well suited to the establishment of genetic models of SI systems, as described above. However, for the next step, that is identification of S gene products at the protein level, only Brassica and Nicotiana species were used during the 1970s, because the protein products derived from their female S genes are highly expressed in stigma/style tissues. Thus, the SI researchers who selected these species for study had a great advantage for molecular dissection of the S-locus.¹⁾

In the case of the Brassica species, immunological cross-reactivity and isoelectric focusing (IEF) were applied to identify the allelic diversity of the S gene products.^16,17) These two methodologies had advantages for the identification of genetic diversity of multi-allelic gene products. For example, in the case of IEF, the molecular masses of allelic gene products are similar. However, they differ in the isoelectric point (pI) because the amino acid composition of their products must be different as a consequence of the allelic diversity and tertiary structure of the proteins. Because the S-locus-specific protein was found to react with concanavalin A (Con A), which is plant lectin,¹⁸⁾ it was concluded that this protein has a carbohydrate moiety, and it was named S-locus glycoprotein (SLG).¹⁾

In the 1980s, SLG proteins were extracted and purified from plant material. Furthermore, the cDNA clones encoding SLG genes were obtained from Brassica rapa (syn. campestris) and B. oleracea.^19,20) At this time, the SLG was proposed to be the female S gene based on several characteristics, as described below. Firstly, the glycoprotein is directed to be secreted into the cell wall by a hydrophobic signal peptide.^19–21) Secondly, twelve cysteine residues were found to be completely conserved in the C-terminal region and variable regions were found in the central and C-terminal regions, from comparisons of over 30 SLGs derived from different S alleles (Fig. 5). When the nucleotide sequences of allelic SLGs were aligned, three highly divergent regions were identified and other regions were found to be highly conserved among S alleles.⁵⁾ These variable regions were proposed to contribute to S allelic specificity, and could interact with the male S determinant. Thirdly, the glycosyl residues were similar among S alleles, indicating that the S allelic specificity is due to variable regions of SLG itself.^1,20) Fourthly, identification of the S allele by PCR methods using primers positioned in the conserved regions has contributed to the breeding of Brassica F₁ hybrid varieties.^10,22,23) However, this method sometimes resulted in the false amplification of the many other SLG-like genes in the Brassicaceae genome.^24–27) To exclude non-S-locus SLG-like sequences, linkage to the S locus had to be confirmed by test pollinations using a Brassica breeding system.⁵⁾

Figure 5.

Schematic structure of SLG and SRK. SLG is an S allele-specific secreted glycoprotein, having a hydrophobic signal peptide in the N-terminus and 12 cysteine residues at the C-terminus. SRK is a receptor protein kinase, whose extracellular domain (receptor domain) is similar to SLG. SRK consists of a signal peptide, an SLG-like domain (S-domain), a hydrophobic transmembrane domain and a cytoplasmic catalytic domain (kinase domain) of the serine/threonine protein kinase type.

Furthermore, in the 1980s, to construct cDNA libraries and determine the amino acid sequence of SLG, tens to hundreds of thousands of stigmas, the top part of the pistil, were necessary, because SLG expression is spatiotemporally controlled in mature stigma papilla cells.^21,28) Thus, this identification of SLG residing at the S locus gave a great advantage to Brassica SI researchers in the further molecular dissection of the S locus, and subsequently, during the 1990s, in determining the sets of male and female SI genes, as described below. However, at that time, comparisons of SLG to sequences in public databases did not provide any evidence as to the function of SLG in SI recognition.⁵⁾

3. Identification and characterization of female S determinants

Following SLG identification, molecular dissection of the S locus was advanced by several important findings. One of the most important milestones was the discovery of ZmPK1, encoding receptor-like protein kinase of maize, whose receptor domain has homology to SLG.²⁹⁾ After this finding, several allelic SRK genes, encoding S-locus receptor-like kinases, located at the S locus, were isolated and characterized from B. rapa, B. oleracea and B. napus. SRK contains three functional domains: an extracellular S domain (receptor domain), a transmembrane domain, and an internal serine/threonine protein kinase domain (Fig. 5).^28,30–32) Interestingly, sequence similarity at the amino acid level of SLG and the S domain of SRK within S alleles is over 80%,⁵⁾ indicating that these two regions arose by duplication within the S locus.^28,33)

Thus, at that time, two stigma-specific genes, SLG and SRK, were known to be present at the S locus. However, it was difficult to determine the physical distance between the SLG and SRK genes precisely. In the case of B. oleracea, the size of the S locus was estimated to be 350 kb by PFGE analysis in the early 1990s,³⁴⁾ although this depended on the location of recognition sites of restriction enzymes around the S locus. An S-locus region as large as 350 kb suggested that cloning and sequencing of the S-locus might be quite difficult. However, this was overcome by the second important breakthrough, the successful cloning of the 76-kb genomic fragment containing both SLG and SRK genes in the S⁹ allele of B. rapa by use of the PAC vector.³⁵⁾ Based on this success, molecular dissection of the whole S locus was possible at the nucleotide sequence level. B. rapa had an advantage over B. oleracea for molecular dissection of the S-locus region because the size of the S locus of B. rapa was smaller than that of B. oleracea.³⁶⁾ Thus, the physical genomic structure of the S-locus region was determined in B. rapa, and also in B. napus (Fig. 6).^37–40) Interestingly, in the case of B. napus, the S-locus region was found to have originated from the introgressed B. rapa genome and not from B. oleracea.^31,37)

Figure 6.

Physical map of the S locus of the S⁹ line of B. rapa, showing the location of 14 genes, a transposon-like sequence and three open reading frames (ORFs). Filled boxes indicate locations of the 14 genes, and horizontal arrows above the genes indicate directions of their transcription. Three long ORFs (a, b and c) and a transposon-like sequence are represented by open boxes labeled “a,” “b,” “c” and “Transposon.” Three genes represented by orange-colored boxed are specifically expressed in stigma tissues. Two genes represented by purple-colored boxes are specifically expressed in anther tissues. The other nine genes represented by green-boxes are expressed in both vegetative and reproductive tissues.

Next, it was important to determine whether SLG and/or SRK are necessary for SI recognition at the stigma side, and for this several gene transformation studies were performed in the 1990s. Loss-of-function experiments showed that SLG and/or SRK are important for the SI recognition reaction.^5,6,41–43) In most cases, following introduction of SLG or SRK a change to the self-compatibility (SC) phenotype was observed, which might have been induced by co-suppression as shown by the change in petunia flower color.^6,44) In order to overcome co-suppression between endogenous and exogenous SLG and/or SRK genes, Takasaki and co-workers used lower-homology SLG/SRK genes from the two groups of S alleles, class I and class II.⁴⁵⁾ As described below, S alleles (S haplotypes) of class I (e.g. S²⁸) are dominant over those of class II (e.g. S⁶⁰) at the pollen side.^10,32) Takasaki and co-workers introduced SLG²⁸ and SRK²⁸ into recipient plant lines, S⁶⁰ homozygote or S²⁸S⁶⁰ heterozygote of B. rapa. As a result, the transgenic plants harboring SLG²⁸ did not show a change in S phenotype at the stigma side, while the transgenic plants harboring SRK²⁸ showed the S allele specificity at the stigma side, but not in pollen. Interestingly, transgenic plants harboring both SLG²⁸ and SRK²⁸ showed the rigid SI phenotype comparable to non-transgenic S²⁸S⁶⁰ heterozygotes, indicating that SRK is the female S determinant and SLG is required for full manifestation of the SI response (Fig. 7).⁴⁵⁾ This additional effect of SLG was not observed in B. napus;⁴⁶⁾ the difference in SLG effect may be a consequence of the genome organization of the diploid (B. rapa) and amphidiploid (B. napus) species.

Figure 7.

Test pollinations for evaluation of SLG and SRK transgenic plants by seed formation. Line S⁶⁰ was used as the recipient of the transgene and line S²⁸S⁶⁰ was used as the pollen tester. In this case, the S phenotype of all pollen was S²⁸ because S²⁸ is dominant over S⁶⁰. The phenotype of transformants carrying SLG²⁸ was coincident to that of the non-transformant recipient plant, that is S⁶⁰. In contrast, the seed formation of transformants having SRK²⁸ decreased compared to that of transformants having SLG²⁸ and recipient S⁶⁰ line. After genetic crossing of the two transformants, which carried SLG²⁸ and SRK²⁸ transgenes, the F₁ plant showed a rigid SI phenotype, like the S⁶⁰S²⁸ heterozygote. SI phenotype was assessed by the number of seeds formed per pod.

4. Identification and characterization of male S determinants

In the course of investigating the identification of the male S determinant, several S-linked genes were found to be unrelated to the male S determinant.^47–50) However, another detailed approach, analysis of pollen coat proteins (PCPs), successfully lead to identification of the male S determinant. In Brassica species, PCPs, which are derived from anther tapetum, are essential for pollen development.⁵¹⁾ Initially, SLG-interactive PCPs were searched,⁵²⁾ and these interactive molecules were then characterized as cysteine-rich small proteins.^53–57) In the course of these experiments, a bioassay system showed that a <10 kDa PCP is the male S determinant.⁵⁸⁾ Interestingly, a gene encoding a PCP-like protein, SP11, was found to be located at the S locus, near SRK in B. rapa; thus SP11 was the most likely candidate for the male S determinant (Fig. 6).³⁸⁾ Subsequently, it was demonstrated that SP11 is the male S determinant by using bioassay and transgenic experiments in B. rapa.^59,60) Around the same time, a gene termed SCR (S-locus cysteine-rich protein) in B. oleracea was also cloned, and identified as the male S determinant.⁶¹⁾ Because SP11 and SCR are identical,^38,59,61) SP11 will be used as the name of the male S determinant in this review article, in both Brassica and Arabidopsis species.

After identification of SLG, SRK and SP11 genes, many allelic genes were isolated,^5,6) which made it possible for us to calculate the allelic diversity. As described above, S alleles were divided into two groups, class I and class II, based on the sequence similarity between allelic genes of the three S-locus genes, SLG, SRK and SP11. In the case of SLG and SRK, within a group, allelic diversity was from a few percent to around 20%. In contrast, between groups, the diversity was over 30%. In contrast, the allelic diversity of SP11 was higher than that of SLG and SRK.^5,6) Particularly in the class I alleles of SP11, only certain amino acid residues (glycine, aromatic amino acid and cysteine residues) were conserved in the mature protein.²³⁾ From determination of the tertiary structure with 2-D NMR analysis, these amino acids were found to be important for recognition specificity.⁶²⁾ From comparison of SP11 and SRK sequences, these two genes encoding the pollen ligand and stigma receptor, respectively, were found to have co-evolved in a trans-specific mode.^6,23,39)

5. Molecular mechanisms of the SI reaction

After identification of both male and female S determinants, the next research targets were the physical interaction between SP11 and SRK, and the SRK-related signaling cascade following this interaction. For the physical interaction between SP11 and SRK, different results have been reported by two laboratories. Shimosato et al.⁶³⁾ showed that both the S-domain and membrane-anchoring domain were necessary for allele-specific interaction by using cross-linking and immunological methods. In contrast, Kachroo et al.⁶⁴⁾ showed that only the S-domain of SRK could interact with SP11 in the allele-specific manner by using an immunoprecipitation method. In the future, after determination of the tertiary structure of SRK, this inconsistency is likely to be resolved.

The physical interaction of SP11 and SRK was found to trigger autophosphorylation of SRK in an allele-specific manner.⁶⁵⁾ The question was then, how does the SI signal transduce in the cytoplasm? In order to clarify the SRK signaling cascade, several SRK-interacting molecules were identified and characterized.^66–72) MLPK (M locus protein kinase), identified from the self-compatible line yellow sarson, was found to encode a novel membrane-anchored cytoplasmic protein kinase,⁷¹⁾ which interacts with SRK directly.⁷²⁾ ARC1 (arm repeat containing 1) and THL1 (thioredoxin 1) were isolated using a yeast two-hybrid system with the kinase domain of SRK as a bait.⁶⁶⁾ ARC1 interacts with the kinase domain of SRK in a phosphorylation-dependent manner and has E3 ubiquitin ligase activity, indicating that the proteasome protein degradation system is involved in SI signal transduction. Furthermore, an ARC1 knock-out line showed the self-compatible phenotype.^67,69) At present, only MLPK and ARC1 have been identified as positive effectors of the Brassica SI reaction. Other identified factors found to date may be negative effectors of SI.^68,71) Recently, novel self-compatible lines of B. rapa have been identified, which will enable progress to be made in understanding the role of other interactive molecules in the complete SI signaling cascade.⁷³⁾ A schematic model of the current understanding of Brassica SI is shown in Fig. 8.

Figure 8.

Schematic model for self-pollen recognition in Brassica species. Male and female S determinant genes, SP11 and SRK, are located at the S locus. SP11 is predominantly expressed in the tapetum cells of anther locules, and accumulates on the pollen surface during pollen maturation. On self-pollination, SP11 molecules penetrate into the papilla cell wall, and interact with SRK in an S-allele-specific manner. Phosphorylated SRK interacts with MLPK. After the subsequent signal transduction, which has not yet been determined, rejection of the self-pollen occurs.

Other approaches to studying the SI signaling cascade have used Arabidopsis species. Because Arabidopsis thaliana is a self-compatible model plant, a very large number of traits of A. thaliana have been studied.^74–77) In addition, for research into the evolution of SI and SC, Arabidopsis species are powerful research tools.^78–82) As describe above, A. thaliana is a model cruciferous plant, which naturally has an SC phenotype, whereas other species, such as A. halleri and A. lyrata, are self-incompatible. A. thaliana modified to be self-incompatible is a powerful tool for genetic approaches to studying the SI signaling cascade, because the whole genome sequence of A. thaliana has been determined²⁶⁾ and other genetic tools (e.g. T-DNA tag lines, many ecotypes and microarray expression data) are available (http://www.arabidopsis.org/). To produce self-incompatible A. thaliana by transformation, two approaches have been used. Nasrallah and co-workers introduced SP11 and SRK from self-incompatible A. lyrata into self-compatible A. thaliana (C24 ecotype) and the resulting transgenic A. thaliana, with both SP11 and SRK, exhibited the SI trait.^83,84) On the other hand, Tsuchimatsu, Suwabe and co-workers introduced only SP11 into the Wei-1 ecotype of A. thaliana, which has a functional SRK but shows SC, to produce a self-incompatible A. thaliana.⁸⁵⁾ In the course of sequencing analysis of the S-locus genes in self-incompatible A. halleri, the SP11 gene of haplogroup A of A. halleri was found to be quite similar to the disrupted SP11 gene of haplogroup A of A. thaliana; it is suggested that a 213-bp inversion of SP11 might have occurred in A. thaliana. Furthermore, from test-crosses using the pollen of the haplogroup A of A. halleri, it was found that several ecotypes (e.g. Wei-1) were functional on the female side of the SI system, indicating that these ecotypes may have a functional SRK and SI signaling cascade. By using Wei-1 as a representative of these ecotypes, self-incompatible transgenic A. thaliana plants carrying the restored SP11 gene were successfully produced (Fig. 9).⁸⁵⁾ This transgenic A. thaliana plant with the restored SP11 transgene, in which the recipient and donor are of the same origin, is an impressive achievement because this transformation represents the artificial retrograde evolution of the SI gene.

Figure 9.

Restoration to SI phenotype in Arabidopsis thaliana. The Wei-1 ecotype usually has a self-compatible phenotype (formation of large siliques and many seeds). The reason for the SC phenotype Wei-1 is a gene inversion within the 2nd exon of SP11, although SRK is functional. When transformed to restore an active SP11, the transformants showed the self-incompatible phenotype (formation of small siliques and no seeds).

6. Molecular analysis of dominance relationships between S alleles

An important characteristic of Brassica SI is the dominance relationships between S alleles, which are a consequence of the sporophytic behavior of S genes.^9–13) Genetic experiments have revealed four characteristics of the dominance relationships of cruciferous plants:^9–13) (1) co-dominance is more frequent than dominance/recessiveness; (2) dominance/recessiveness in the pollen is observed more frequently than that in the stigma; (3) dominance relationships act independently in the pollen and stigma; and (4) non-linear dominance relationships are also observed, and are more frequent in the stigma than in the pollen. Discovery of the molecular mechanisms of dominance relationships has been based on these genetic features.

In Brassica pollen, dominance relationships of SI are regulated by transcription of SP11. In sporophytic tapetum cells of anthers of S heterozygous plants, SP11 derived from the dominant allele is normally expressed, whereas expression of the recessive SP11 is significantly suppressed (Fig. 10A).^86,87) Interestingly, linear dominance relationships (S⁹ > S⁴⁴ > S⁶⁰ > S⁴⁰ > S²⁹) are also observed in B. rapa.^10,87) In this case, dominance/recessiveness of S⁴⁴, S⁶⁰, S⁴⁰ alleles could be altered, indicating involvement of epigenetic regulation.⁸⁷⁾ From observation of the methylation level of the promoter region of SP11 in several S heterozygotes, it was shown that the recessive SP11 is specifically methylated in S heterozygotes.⁸⁸⁾ Recently, it was further demonstrated that the small RNA produced from the dominant allele could activate methylation of the recessive allele. In transgenic experiments, S⁶⁰ (class II) transformants with a class-I-derived small RNA region (S⁹) showed SC, and their promoter regions were highly methylated, as in the S⁹S⁶⁰ heterozygote, indicating that the small RNA from the dominant allele functions in trans to induce transcriptional silencing of the recessive allele (Fig. 10B).⁸⁹⁾

Figure 10.

Schematic model for molecular mechanisms of dominance relationships at the pollen side. (A) In the case where S¹ is dominant over S², dominant transcripts of SP11, S¹-SP11, are specifically expressed in the S¹S² heterozygote. However, S²-SP11 transcripts are not detected in the S¹S² heterozygote on RNA gel blot analysis. The results demonstrate that the dominance relationship at the pollen side is regulated at the transcriptional level. (B) In the dominant S allele, small RNA, termed Smi (SP11 methylation inducer), is specifically produced, and its nucleotide sequence is highly similar to the promoter region of the recessive SP11 gene. This small RNA induces the methylation of recessive SP11, and represses the recessive SP11 at the transcriptional level.

In contrast, dominance relationships of SI at the female side appear to be post-transcriptionally regulated by SRK, unlike transcriptional suppression of SP11.⁹⁰⁾ The different mechanisms operating in the male and female side in the dominance relationships of SI are consistent with the four genetic characteristics described above.^9–13)

7. Future prospects

As outlined in this review, the Brassica SI reaction has several interesting biological features, including cell-cell communication, ligand-receptor interaction, signal transduction, phosphorylation cascade, molecular evolution, allelic polymorphism and epigenetic regulation by small RNAs. For an overall understanding of SI events, collaborative research with diverse biological approaches (metabolome, proteome, bioinformatics, phenome, transcriptome, etc.) is necessary (e.g. refs. 91–119).

SI is one of the most interesting phenomena in sexual reproduction. Building on recent progress in sexual plant reproduction research, the molecular processes in male and female gametophyte development have now been extensively dissected in several plant species (e.g. refs. 104, 118–136). Such accumulation of biological knowledge will also contribute to the understand of pollen germination and pollen tube behavior in the SI recognition reaction.

Interestingly, peptide signals similar to SP11 are common in several other related phenomena (fertilization, morphogenesis, plant-microbe interaction, etc.; e.g. refs. 15, 137–141). To elucidate a complete overview of the peptide signaling, further analysis of protein-protein interaction by use of the yeast two-hybrid system, protein cross-linking, etc. is required,^142,143) in addition to the genetic analysis described above. To date, binding of SP11 to the SRK receptor, the tertiary structure of SP11 and the variable regions of SP11 have been determined.⁶²⁾ However, the tertiary structure of SRK has not yet been established. When this has been carried out, the precise SP11-SRK allelic-specific interaction will provide new insight into SP11-induced SRK activation, as has been achieved for brassinosteroid and BRI1.¹⁴⁴⁾

As the next important topics for study in pollen-stigma interaction, genetic barriers between different species and interspecific incompatibility are interesting phenomena, and these are also related to the SI reaction. Although genetic analysis of interspecific incompatibility is complicated due to the difficulty or impossibility of obtaining hybrid seeds and their progenies,¹⁾ the understanding of molecular mechanisms in the SI recognition reaction will help to identify the molecular players in interspecific incompatibility. In addition, overcoming the barrier of interspecific incompatibility could lead to the establishment of new species. From observations of chromosome hybridization, it was determined that the amphidiploid B. napus originated from crosses between B. rapa and B. oleracea.¹⁴⁵⁾ Interestingly, the parental diploids, B. rapa and B. oleracea, show SI phenotype and their amphidiploid, B. napus, has the SC phenotype. In some SC B. napus, the mutated genes conferring SC phenotype have been identified and characterized.^146,147) Taken together, these points indicate that further studies of these Brassica SI and SC lines should contribute to our understanding of the molecular mechanism of evolution from SI to SC during polyploidy formation under cultivation.

Very many reports have been published of the early physiological studies.⁵¹⁾ In Brassica, breakdown of SI was reported to be caused by several treatments, including CO₂ gas, NaCl solution, high temperature, organic solvents and electrical stimulation.^148–154) However, the molecular mechanisms involved in overcoming the SI recognition reaction are still unknown. Understanding the mechanisms in overcoming SI will contribute to the discovery of the missing links in our understanding of the physiological aspect of the Brassica SI reaction. Another factor is the cuticle layer of stigma papilla cells, which is important for Brassica SI. Even in an incompatible cross the pollen tube can penetrate into the cuticle layer,¹⁵⁵⁾ indicating that there is ‘cutinase’ activity on the surface of pollen grains. Recently, results of studies on cuticle wax in other tissues have been published,^156,157) and these are informative for dissecting Brassica SI. Thus, from earlier studies we can obtain many pointers to advance Brassica SI research in the future.

Finally, just as we were starting to write this review article, the huge earthquake (magnitude 9.0) hit the northern part of Japan, around Sendai, on March 11, 2011.¹⁵⁸⁾ The Fukushima Daiichi nuclear power station was severely damaged by the huge earthquake and tsunami, and radio-active elements (iodine-131, cesium-137 and other compounds) were released into the environment.¹⁵⁹⁾ For the remediation of soil contaminated by radioactive compounds, especially cesium-137, Brassica species are potentially very useful.^160–165) Of course, Brassica crops, cabbage, turnip, broccoli, etc., are important as food crops for people in all parts of the world. In addition, the breeding of Brassica crops has the potential to contribute to production of bio-fuel, phytoremediation and phytomining.¹⁶⁶⁾ In view of these points, understanding molecular mechanism of the SI recognition reaction in Brassica species will be important in establishing F₁ hybrid seed production in the future.

Biographies

Profile

Masao Watanabe was born in Ehime Prefecture in 1965, and received his Ph.D. in 1994 from Tohoku University under the direction of Prof. Kokichi Hinata. He was an Assistant Professor at Tohoku University (1991–1997), Associate Professor at Iwate University (1997–2005). Since 2005, he has been a Professor of Plant Reproductive Genetics, Tohoku University. His research interest is the molecular dissection of self-incompatibility in Brassica species. He received JSPS PRIZE in 2011.

Keita Suwabe was born in Kanagawa Prefecture in 1974, and received his Ph.D. in 2004 from Mie University under the direction of Prof. Yasuo Kowyama. Since 2009, he has been an Associate Professor, Mie University. His research interest is the molecular dissection of pollen-stigma interaction in higher plants. He received Japanese Society of Breeding Award for Young Scientist in 2012.

Go Suzuki was born in Shizuoka Prefecture in 1971, and received his Ph.D. in 1999 from Tohoku University under the direction of Prof. Kokichi Hinata. He was an Assistant Professor at Osaka Kyoiku University (1999–2003). Since 2003, he has been an Associate Professor, Osaka Kyoiku University. His research interest is the functional analysis of genomic organization in higher plant genes. He received Japanese Society of Breeding Award for Best Paper in 2005.