Over the past decade or so, there has been significant progress towards elucidating the molecular events occurring during pollination in flowering plants. This process involves a series of complex cellular interactions that culminates in the fusion between male and female gametes. The process also regulates crucial events such as pollen adhesion, hydration, pollen tube growth and guidance to the ovules. Additionally, in many instances, incompatibility mechanisms that control the acceptance or rejection of pollen alighting on a recipient plant play a major role in the pollination process. In this article we aim to review our current understanding of the components that are implicated in enabling the pollen to deliver the male gametes to the ovary and the molecular mechanisms by which they are thought to act.
IV. Pollen germination and initial growth on the stigma surface 568
V. Pollen tube growth through the style and pollen tube guidance 569
VI. Control of pollen viability by incompatibility responses 572
1. Self incompatibility (SI) 573
Gametophytic SI 573
SI in the Solanaceae 573
SI in Papaver 575
Sporophytic SI 577
SI in Brassica 577
SI in Ipomoea 579
2. Interspecific incompatibility responses 579
VII. Conclusions and perspective 580
Reproduction is a crucially important stage in the life cycle of all organisms. As a consequence, there is an intense interest in determining how this is controlled. The central importance of reproduction in flowering plants has provided the justification and driving force behind the extensive studies that have focused on establishing the fundamental basis of this highly complex process. Moreover, developments in molecular biology and genetics will, no doubt, enable future applications based on these fundamental studies, perhaps permitting the manipulation of reproductive processes, which could potentially have uses for plant breeding.
In flowering plants one of the most important steps in the reproductive process is pollination. This has been the subject of studies stretching back several centuries to pioneers such as Joseph Kolreuter who, in the 18th century, carried out extensive studies of hybridizations between plant species. Later, Charles Darwin was both fascinated and perplexed when noting that, in some instances, species with completely normal reproductive organs were unable to self-pollinate, yet were perfectly capable of pollinating other individuals (Darwin, 1877). These proved to be the earliest observations of self-incompatibility. In this article we discuss the recent progress in elucidating the events that follow the arrival of a pollen grain on a receptive stigma. A cartoon of the route of the pollen from landing on the stigma, and its journey through the pistil to the ovary is shown in Fig. 1, together with a summary of some of the components identified as being involved in pollination, and discussed in this review. We also discuss the role of incompatibility systems, that act in a significant number of species to control the acceptance or rejection of pollen alighting on a recipient plant, hence, playing a major role in the pollination process.
II. Adhesion of pollen to the stigma
Initiation of pollination is dependent on the ability of the pollen grain to adhere effectively to the stigmatic surface. The stigmatic surface of different plant species varies widely in both morphology and the presence or absence of exudates, and these are likely determinants of the relative importance of adhesion in any given species (Heslop-Harrison & Shivanna, 1977). Control of pollen acceptance by adhesion is seen as particularly important in species, such as those in the Brassicaceae, that have a dry stigma. By contrast, the surfaces of species, such as those belonging to Solanaceae and Leguminosae that have wet stigmas appear to promote the adhesion of most pollen species (Zinkl & Preuss, 2000). Studies by Preuss et al. (1993) indicate that adhesion is under polygenic control. It is also apparent that, pollen adhesion in Brassica spp. increases over time, requiring 30 min for maximum binding (Heizmann et al., 2000). These two pieces of evidence suggest that the control of pollen adhesion is likely to be complex (see Fig. 1).
In the Brassicaceae it is known that the pollen coat or tryphine contains components involved in mediating cell–cell interactions between pollen and stigma (Doughty et al., 1993; Preuss et al., 1993; Dickinson, 1995). The presence of the pollen coat is essential for attachment of pollen to the stigmatic surface (Stead et al., 1980; Elleman & Dickinson, 1986). It is also apparent that the application of isolated pollen coating invokes physiological changes to the stigmatic papillae, notably a rapid expansion of the outer layer of the stigmatic wall (Elleman & Dickinson, 1996). The interaction between stigma and pollen also results in changes in the pollen coat enabling hydraulic continuity between the pollen grain and the stigmatic papillae (Elleman & Dickinson, 1986).
Most of the current research on pollen adhesion is focused on factors affecting this process in Arabidopsis thaliana, Brassicaoleracea and B. rapa. Interactions between stigma and pollen have been studied using either centrifugation (Luu et al., 1997), detergent assays or spring displacement experiments (Zinkl et al., 1999). It has been demonstrated that A. thaliana stigmas selectively bind A. thaliana pollen with much higher affinity than pollen from related species and that this interaction occurs within seconds of pollination (Zinkl et al., 1999). In these studies, pollen from a range of monocotyledenous species was found to have little or no binding capacity to A. thaliana stigmas. Also, whilst pollen from several dicotyledonous species exhibited a range of binding capacities, in all instances it could be washed off by a detergent treatment that had little effect on A. thaliana pollen binding (Zinkl & Preuss, 2000). Interestingly, the study included Brassica campestris, which like A. thaliana is a member of the Brassicaceae. Despite this close relationship, the behaviour of B. campestris pollen binding was no different to that of other dicotyledonous species, indicating a considerable degree of specificity in this event.
A genetic screen using male sterile pistils, with the aim of isolating mutants showing reduced pollen adhesion (Zinkl & Preuss, 2000), has resulted in the isolation of several lap (less adherent pollen) mutants. The mutant lap1 shows gross defects in the exine, suggesting that it is the pollen coat that is important in pollen adhesion. The fact that this mutant does not exhibit reduced fertility reveals that the adhesion process is independent of pollen hydration and growth. As yet, no genes have been linked to the lap mutants.
Two stigmatic proteins have been implicated in playing a role in adhesion in Brassica. A recent study of two members of the Brassica self-incompatibility S gene family (see later), the S-locus glycoprotein (SLG) and the S-locus related protein (SLR1) has suggested that they may both be involved in this process. The force of adhesion between Brassica pollen and stigmas has been measured using a centrifugation assay (Luu et al., 1997). In transgenic plants in which SLR1 expression was down-regulated by antisense suppression, a reduction in adhesive force was found, whilst there was no effect on the ability of the stigma to support pollen tube growth. A similar result was obtained when wild-type stigmas were treated with either anti-SLR1 or anti-SLG antibodies (Luu et al., 1999). The data from Luu et al. (1999), together with the localization of SLG and SLR1 within the cell wall rather than the pellicle (Kandasamy et al., 1989; Umbach et al., 1990), suggest that these proteins are involved in later steps of adhesion, rather than the initial binding event.
Both SLG and SLR1 have been shown to interact in vitro with pollen coat proteins (PCPs) (Doughty et al., 1993; Hiscock et al., 1995). The PCPs are an extensive family of gametophytically expressed, small, cysteine-rich proteins. Gel shift assays have revealed that two family members PCP-A1 and PCP-A2, interact with SLG and SLR1, respectively (Doughty et al., 1993, 1998; Hiscock et al., 1995). It is well established that, on contact with the stigmatic surface, the pollen coat flows from between the baculae of the sporopollenin exine to form an adhesive foot at the papilla surface (Stead et al., 1979; Elleman & Dickinson, 1990). Thus, it is reasonable to assume that this event would facilitate the PCP-SLR/SLG interaction in vivo (Doughty et al., 2000). Initially, it had been suggested that the PCPs might be involved in some way in the Brassica incompatibility systems, which will be discussed later (Dickinson et al., 1997; Doughty et al., 2000). Whilst their exact function has not yet been unequivocally established, at least some members of the PCP family may play a role in pollen adhesion through their interaction with SLR1/SLG.
III. Pollen hydration
Following adhesion of pollen to the stigmatic surface, successful pollen tube growth depends upon the hydration of the pollen grain (see Fig. 1). The processes involved in pollen hydration are, however, not as well characterized as those involved in adhesion. This, in part, is due to technical difficulties associated with the study of this process. The diverse nature of the stigmatic surface in different angiosperm families suggests that hydration mechanisms are also likely to be divergent, although the results of hydration, notably a far-reaching reorganization of the vegetative cell (Heslop-Harrison & Heslop-Harrison, 1992), are likely to be similar.
There is considerable evidence to suggest that long-chain lipids act as signals to stimulate pollen hydration. Mutants with defects in pollen hydration, such as the cer mutants and pop1, have been shown to have defects in lipid biosynthesis (Preuss et al., 1993; Hülskamp et al., 1995a). Studies by Wolters-Arts et al. (1998) demonstrated that lipid-rich stigma exudates from Petunia could restore normal pollen germination and growth in transgenic Nicotiana tabacum, in which the secretory zone of stigmas was ablated using a cytotoxic STIG1-barnase gene (Goldman et al., 1994). Application of one of the exudate lipids, the long chain lipid trilinolein, enabled pop1 mutants to germinate normally. It also enabled pollen to hydrate and germinate on leaf tissue following removal of the cuticle, which is normally a nonreceptive surface (Wolters-Arts et al., 1998). Further support for the role of the exudate in hydration was obtained in studies of Nicotiana, during which it was observed that pollen grains alighting at the top of stigmatic papillae, that is above the level of the exudate, failed to hydrate (Lush et al., 1998).
Although the evidence indicates a key function for lipids as signals for pollen hydration, it now seems likely that they are not the only molecules involved in this process. Recently, an A. thaliana mutation that results in the loss of the oleosin-domain GRP17, the most abundant pollen coat protein, has been described (Mayfield & Preuss, 2000). As a result of this mutation, pollen hydration was delayed almost three-fold compared to wild-type pollen, which normally hydrates within 4–5 min of alighting on the stigma. The delay was due to a failure to interact with the stigma, rather than failure to absorb water per se, as once hydration had initiated, the mutant pollen expanded at the same rate as wild-type pollen. Although the grp17 pollen does eventually hydrate, the authors speculate that the mutation substantially reduces its fitness compared with wild-type pollen. Whether GRP17 directly communicates with the stigma or modulates the activity of other molecules, such as lipids, remains to be established (Mayfield & Preuss, 2000).
One major breakthrough in our understanding of components that potentially regulate water flow to pollen grains during hydration, was the identification of an aquaporin that is essential for pollen hydration (Ikeda et al., 1997). Aquaporins are water channel proteins (Chrispeels et al., 1999). They are abundant components of the plasma membrane existing as a tetramer of 27 kDa subunits, each comprised of six membrane-spanning domains. It is thought that each individual polypeptide forms a water channel, rather than the channel being formed by the tetrameric structure. Although the aquaporin described by Ikeda et al. (1997) has been specifically implicated in the control of self-incompatibility (see later), it seems apparent that hydration in other species is likely to depend upon the presence of similar proteins.
IV. Pollen germination and initial growth on the stigma surface
Once pollen is correctly hydrated it must then germinate. It will then grow through the stigma before it reaches the stylar tissues and eventually the ovules (see Fig. 1). The initial penetration of the stigmatic surface has been studied in several species with dry stigmas and appears to be quite variable. In A. thaliana and B. oleracea the pollen tube penetrates the stigmatic cuticle and enters a space between the outer layer of the cuticle and the main body of the fibrillar cell wall, where it continues to grow until it reaches the base of the cell. It then enters the stigma transmitting tissue where it grows intercellularly. In Papaver rhoeas the pollen tube grows underneath the cuticle to the base of the papillar cell. In the Asteraceae pollen tubes appear to grow extracellularly until they reach the base of the papillae, although as they reach the base they then grow through the middle lamella as is the case in the Brassicaceae (Elleman et al., 1992).
In the Solanaceae, it has been proposed that lipids direct pollen tube growth by controlling water flow to pollen (Wolters-Arts et al., 1998). The major components of the hydrophobic exudate in the Solanaceae are triglycerides (fatty acid chains – C18) (Cresti et al., 1986). It is postulated that the hydrophobicity of the exudate is a critical factor in the ability of pollen to penetrate the stigma (Lush et al., 2000). This establishes a gradient of water in the exudate that is used as a guidance cue by the germinating pollen tube (Lush et al., 1998). Lush and co-workers have attempted to reconstruct the stigmatic environment in vitro (Lush et al., 2000). This work is based around the use of a two-phase system comprising a hydrophobic phase, representing the exudate, and a hydrophilic phase, the equivalent of the stigma cells and aqueous drops on the stigma surface. The hydrophobic phase was created by the use of exudate or specific oils, whilst the aqueous phase was derived from the pollen growth media. These studies revealed that the speed of pollen hydration and germination is related to the proximity between the aqueous phase and pollen grains suspended in the exudate/oil phase. The pollen tube emerged from the aperture closest to the aqueous phase and was directed towards this phase. This suggests that polarity of pollen tube growth may be determined by this simple physical cue. Such a guidance mechanism that depends on the physical environment in the stigma is independent of pistil components. This suggestion is supported by the observation that several oils were an effective substitute for stigmatic exudate. This proposal, that the physical environment in the pistil guides pollen tube growth, was extremely contentious when first proposed because it appears to contradict the widely accepted view that the specific chemical environment provides guidance cues.
Pollen tubes enter the stigma by growing through the intercellular spaces (Cresti et al., 1986), and although they do not appear to be guided by high precision, their growth along the surface probably provides a physical cue that increases the likelihood of penetration occurring (Lush et al., 2000). Mechanical cues for signalling have been shown to be used by fungal hyphae, which, like pollen, also elongate by tip growth. They have been shown to employ topographical guidance systems to determine where their appressoria form, in a thigmotropic response. This has been established by using artificial, microfabricated surfaces to provide specific topographic patterns. Specific spacing and patterning are required for contact sensing, to which the hyphae respond by stimulating differentiation to form infection structures (Allen et al., 1991; Perera et al., 1997; Read et al., 1997). Study of a wide range of fungi suggest that thigmotropism is likely to be a general feature of fungal growth, rather than a specific pathogenic property (Perera et al., 1997). More recently, evidence has been obtained, that suggests this response involves calcium-dependent signalling (Watts et al., 1998). It is possible that this mechanism is involved in other tip-growing cells, such as pollen tubes, though to our knowledge, no one has investigated this possibility.
V. Pollen tube growth through the style and pollen tube guidance
Following penetration of the stigma, pollen tubes grow through the style towards the ovule (see Fig. 1). How an emerging pollen tube is accurately directed towards the ovule is a matter of considerable debate. That some form of guidance or stimulation mechanism is likely to be involved is largely based on observations that in vitro-grown pollen does not generally exhibit any inherent directionality, and the rate of growth is usually significantly slower than that of pollen growing in vivo (Jauh & Lord, 1995). In some species, such as maize and pearl millet, for example (Heslop-Harrison et al., 1985; Heslop-Harrison & Reger, 1988), it was proposed that the directionality of tube growth is due to mechanical influences existing within the stylar tract. In this model of tube growth, the cellular architecture of the transmitting tissue provides the necessary topographic guidance required by the elongating pollen tube to successfully reach the ovary. This type of mechanism has also been suggested for the Solanaceae, where the tubes grow through the intercellular spaces between the parallel files of transmitting tract cells that comprise the rigid, solid styles found in these species (Lush et al., 2000). However, in the case of the Solanaceae, an alternative hypothesis has been proposed based on chemical/biochemical guidance cues within the style that attract pollen tubes and promote growth towards the ovary.
During growth towards the ovary, the pollen tubes are in intimate contact with the components of the extracellular matrix (ECM) of the transmitting tract (Cheung, 1995). The ECM is comprised of a complex mixture of proteins, in particular the arabinogalactan proteins, proline-rich glycoproteins and extensin-like proteins. The arabinogalactan proteins (AGPs) are highly glycosylated, hydroxyproline-rich glycoproteins that are implicated in various aspects of plant growth and development (Baldwin et al., 1993; Lind et al., 1994; Nothnagel, 1997). Studies of pollen tube growth through the transmitting tract of N.tabacum have resulted in the isolation of a class of AGPs, the transmitting-tissue specific (TTS) proteins (Cheung et al., 1993; Wang et al., 1993) that stimulate pollen tube growth in vitro. TTS proteins are encoded by two genes, TTS-1 and TTS-2, that exhibit a high degree of homology. They encode secreted polypeptides that, following cleavage of the signal peptide, have a molecular weight of c. 28 kDa (Cheung et al., 2000). These polypeptide backbones are modified by hydroxylation of the proline-rich domain located towards their N-terminus and extensive glycosylation with N- and O-linked glycans. The most abundant mature TTS protein in N. tabacum has an apparent molecular weight of between 45 and 105 kDa. Interestingly, ectopic expression studies have revealed that glycosylation of TTS is organ-specific, and is restricted to the transmitting tissues of the style (Cheung et al., 1996). The TTS proteins are thought to be multifunctional, in that they are implicated in pollen tube adhesion in the style, pollen guidance and nutrition.
Evidence for the biological function of the TTS proteins was obtained from both in vitro and in vivo studies (Cheung et al., 1995; Wu et al., 1995). Antibodies raised against TTS proteins were used in cyto-immunodetection studies with both in vivo and in vitro grown pollen. These revealed that the TTS proteins adhered to the cell wall and to the tip of the growing pollen tube. In addition, a proportion of the TTS proteins were found to be incorporated into the pollen cell walls, thereby providing a clear indication of the potential for ECM components to influence pollen tube growth. The observation that the addition of TTS proteins to pollen grown in vitro resulted in a threefold growth rate stimulation is also consistent with a direct role in pollen tube growth. Stimulation appeared to be dependent on the presence of glycan moieties, since chemically deglycosylated TTS did not stimulate growth. However, cautious interpretation is warranted since deglycosylation may result in denaturation of the protein. To investigate the role of TTS in vivo Cheung and co-workers constructed transgenic N. tabacum in which the level of TTS protein was reduced, using antisense suppression and sense cosuppression (Cheung et al., 1995). The reduction in TTS protein was found to lead to an overall reduction in seed set, the severity of which was dependent on the degree of TTS down-regulation. This effect was due to a reduced rate of pollen growth rate through the transmitting tissues of the pistil, a finding entirely consistent with the in vitro observations.
These studies additionally provided evidence that suggested that TTS proteins might provide directional cues to extending pollen tubes. Using a semi-in vivo system in which pollen was grown through a cut style placed on growth medium, it was found that on emerging from the base of the cut-style, pollen tubes would grow in the direction of an agarose plug containing freshly prepared TTS proteins. This observation suggested that TTS might be a signal for pollen tube guidance through the style to the ovary. Of course, the difficulty was to explain how it might contribute to directional growth guidance in planta, as TTS is found throughout the extracellular matrix of the stylar canal. One possible explanation, proposed by Cheung and co-workers, is based on their finding that there is a gradient in the level of glycosylation of TTS down the style (Wu et al., 1995). The protein is most highly glycosylated at the ovarian end of the style and this is also reflected in a corresponding increase in acidity. The TTS glycosylation gradient might also be modified through the interaction with the elongating pollen tubes. This is based on two findings: TTS binds to the surface of growing pollen tubes, and possibly acts as an adhesive substrate; and in vitro grown pollen tubes deglycosylate TTS (Wu et al., 1995), suggesting that this might also occur in vivo. It is therefore proposed that pollen tubes grow through the transmitting tissue of the style along the gradient of increased TTS glycosylation. Furthermore, the deglycosylation activity of the growing pollen would contribute by increasing the gradient in the vicinity of the extending tubes and provide a potential nutrient source in the form of the released sugar molecules.
The proposed role of TTS is, however, controversial. Several lines of argument have been raised that question the exact function of TTS in pollen tube guidance. First, the theoretical maximum distance over which chemical cues can guide pollen tubes is between 1.2 and 9.3 mm, compared with the 40–50 mm length of the N. tabacum style (Lush, 1999). Furthermore, even over this shorter distance, gradients of 10 000-fold are required, which is in stark contrast to the fourfold gradient in TTS glycosylation from the top to the bottom of the style (Wu et al., 1995). Earlier studies have also demonstrated that pollen tubes will grow through styles towards the stigma if artificially introduced at the base or the middle of cut styles (Mulcahy & Mulcahy, 1987).
A further question concerning the exact role of TTS arose from the identification of a homologue of TTS, the galactose-rich stylar glycoprotein (GaRSGP) in N. alata (Sommer-Knudsen et al., 1996, 1998). GaRSGP has a polypeptide backbone that is 97% identical to that of TTS. However, it is mostly located within the cell wall, not the intercellular space of the transmitting tract and crucially, unlike TTS, was found to be inactive in pollen tube growth assays (Sommer-Knudsen et al., 1998). This apparent discrepancy has recently been addressed. Stylar transmitting tissue glycoproteins were isolated from N. alata using procedures applied in the isolation of TTS from N. tabacum (Wu et al., 2000), which resulted in the isolation of a set of glycoproteins, referred to as NaTTS proteins that were virtually identical to TTS from N. tabacum. Most importantly, NaTTS proteins were also found to enhance pollen tube growth and to attract tubes in the semi-in vivo assay used previously (Wu et al., 2000). It was suggested that the extraction procedures used in the earlier study (Sommer-Knudsen et al., 1998) had resulted in the isolation of a subset of less heavily glycosylated glycoproteins typified by GaRSGP. Bearing in mind that chemically deglycosylated TTS failed to promote pollen tube growth, it was proposed that the lower level of glycosylation of GaRSGP might explain why it too failed to promote growth (Wu et al., 2000).
Thus, whilst it appears that TTS proteins stimulate growth through the style, whether or not they are providing guidance cues in vivo is less clear. The complexity of the ECM adds to the difficulty in resolving this issue, particularly as other components may also have a function in pollen tube growth. It is likely that there are several different components that interact. In the case of N. tabacum, at least one other class, the class III pistil-specific extensin-like proteins (PELPs), are known to closely interact with the pollen tube as it extends through the stylar canal (de Graaf et al., 1998). Extensins are a class of hydroxyproline-rich glycoproteins found in plant cell walls that are characterized by the presence of numerous repeats of a Ser-(Hyp)4 motif. Studies indicate that the PELPs are actually translocated from the matrix to the pollen tube wall. A similar situation has been described in N. alata where a 120-kDa glycoprotein, which has both extensin-like and arabinogalactan-like properties, has been found to enter the pollen tube during its growth through the style (Lind et al., 1996; Schultz et al., 1997). Similar types of protein have previously been reported in other species. For example, pt11 is a gene expressed specifically in the transmitting tissue of Antirrhinum pistils (Baldwin et al., 1992). Another extensin-like gene was reported in Zea mays, however, in this case, it was a pollen specific extensin-like protein, Pex1, that might potentially recognize and interact with molecules within the style (Rubenstein et al., 1995). These studies suggest that extensin-like proteins may make an important contribution to adhesion between the growing pollen tube wall and the ECM of the pistil transmitting tissue across a wide range of plant families. This could directly influence the efficiency by which the pollen tube negotiates the transmitting tract and facilitate other signalling events between the male and female tissues. These possibilities warrant further investigation.
A number of studies in other plant species have also provided evidence that indicates a role for chemotrophic factors in pollen tube guidance. In particular, they highlight the role of the female gametophyte in a range of species (see Fig. 1). In peach, for instance, secretions from both the ovary and individual ovules appear to have a role in pollen tube guidance and penetration of the ovule (Hererro, 2000). Studies have identified mutants in which guidance of the pollen tube to the ovule is impaired. Wilhelmi & Preuss (1996) screened 80 000 A. thaliana plants before identifying two mutants (pop2 and pop3) that were defective in genes involved in the targeting of pollen tubes to ovules. Instead of growing towards the micropyle and avoiding already fertilized ovules, the pop mutants grew randomly throughout the ovary locule. The pop mutants appear to result from defects in both the pollen and pistil, and the authors noted that there was a failure of pollen tube adherence to the pistil. Also in A. thaliana, ovule mutants have been described which appear to alter pollen tube growth (Hülskamp et al., 1995b). First of all, the growth path of wild-type pollen was examined microscopically. This revealed that in a situation where only limited numbers of pollen tubes grew down the transmitting tract, there was a strong preference to emerge on the surface of the septum at the position of the ovule closest to the stigma. Progressively fewer tubes emerged at ovules located at positions further down into the ovary. This gradient effect was only slightly reduced when the number of pollen tubes growing through the transmitting tract increased.
Ovule selection following emergence of the pollen tube on the septum was also investigated. This revealed that a substantial number of tubes, some 39–46%, grew directly toward the most proximally located ovule. Nevertheless, a significant proportion of the tubes grew toward more distant ovules. When comparable studies were carried out in four ovule mutants (bel1, sin1, 47H4 and 54D12) it was found that the normal behaviour of the pollen tubes was perturbed (Hülskamp et al., 1995b). Emergence of the pollen tube on the surface of the septum no longer favoured the ovule closest to the stigma. In the case of the sin1 and 47H4 mutants in which ovule development was most disrupted, the tubes emerged along the entire length of septum with equal probability. After emerging on the septum surface the ability of the pollen tube to target an ovule was also adversely affected. Instead of directly growing towards a funiculus and the micropyle of a selected ovule, the tubes grew randomly over accessible surfaces. These studies did not identify the exact tissue in the developing ovule that provides the directional signals for pollen tube growth. However, they do provide a link between embryo sac development and pollen tube guidance. In the case of the 54D12 mutants that arrested at various stages in embryo sac development, it was found that in the absence of any development, all pollen tubes failed to find an ovule. In the ovules that arrested at an intermediate stage of development 28% were associated with a pollen tube, compared to the wild-type figure of 92%. It was concluded that the embryo sac, or associated tissues, generates a long-range directional signal that influences the emergence of the tube on to the septum surface, and subsequent guidance to an ovule. This signal did not appear to influence the early stages in pollen tube growth as this was identical in both wild type and mutant plants (Hülskamp et al., 1995b).
Other studies have provided additional evidence in support of an important role for the embryo sac in pollen tube guidance. Ray et al. (1997) used a semisterile A. thaliana line, TL-1, that was heterozygous for a reciprocal chromosomal translocation. This resulted in plants with normal diploid somatic cells. However in this translocation line half of the meiotic products received an abnormal chromosome complement, and consequently did not go on to develop normal embryo sacs. When these plants were pollinated, the pollen tubes were guided to the ovules containing normal embryo sacs but not to those with degenerated embryo sacs. This again leads to the suggestion that viable embryo sacs provide signals to guide the pollen tube. Whether this signal acts directly on the pollen tube or via an influence on the cells in the vicinity of the ovule remains to be resolved (Ray et al., 1997).
The role of the embryo sac in pollen tube guidance has also been investigated in vitro (Higashiyama et al., 1998). In these studies pollen tubes and naked embryo sacs of Torenia fournieri were cocultivated on solid growth media. The pollen tubes were found to grow with a high degree of accuracy toward the micropylar end of the naked embryo sac. This behaviour was completely abolished by heat-treatment of the ovules, providing further evidence in support of a role for proteins produced by the embryo sac in pollen tube guidance. More recently, Shimizu & Okada (2000) have isolated an A. thaliana mutant with an altered phenotype in the female gametophyte. The mutant, magatama (maa), shows delayed female gametophyte development. Pollen tube guidance also seems to be affected by the mutation. Pollen tubes appeared to lose their way just before entering the micropyle, and the mutants also showed a tendency to attract two pollen tubes rather than one. The authors propose a ‘monogamy’ model to account for this, where the female gametophyte emits two attractants (a funiculus guidance cue and a micropyle guidance cue) whilst repulsive forces on the pollen tube prevent polyspermy. Thus, in the case of the maa mutants both the micropyle guidance cue and the repulsive pollen signal are lacking.
The majority of the studies discussed above have focused on the Solanaceae, which have solid stylar transmitting tracts. However, in the case of Lilium longiflorum, the organization of the style is different. The lily style is hollow, with a lining of secretory cells that form the transmitting tract. Based on their studies over a number of years, Lord and co-workers have hypothesized that pollen tube guidance and growth rate through the style is dependent on a haptotactic (matrix adhesion-driven) mechanism (Lord & Sanders, 1992; Lord, 2000). They propose that the pollen tubes adhere to the secretory cells via components in the style ECM and are thereby guided towards the ovary. The influence of the style was suspected from circumstantial evidence based on comparisons of in vivo and in vitro grown pollen. Observations of pollinated styles after cryo-fixation revealed the apparent adhesion of the pollen tubes to the secretory cell lining of the transmitting tract (Jauh & Lord, 1995). By contrast to pollen grown in vitro, pollen tubes growing in vivo adhere to each other, forming star-shaped clusters of F-actin, reminiscent of focal adhesions in moving animal cells. They also grow significantly faster than pollen grown in vitro (Jauh & Lord, 1995, 1996). Furthermore, previous studies had revealed that the transmitting tissue was capable of translocating inert latex beads in a manner reminiscent of pollen tube growth (Sanders & Lord, 1989). These data have given rise to the proposal that pollen tubes are not just tip-growing cells, but act as a moving cell system, similar to neuronal cells, at least if they are in their natural environment, interacting properly with other components in the pistil. This analogy has been explored in a recent review (Palanivelu & Preuss, 2000). These observations prompt the need for more studies of pollen tubes in vivo, rather than in vitro, to obtain a better appreciation of what is really occurring with respect to control of pollen tube growth.
In order to take this model forward, it was necessary to identify components within the style that were responsible for generating adhesion to the growing pollen. Recently, significant progress towards this goal has been attained. An adhesion bioassay, based on the binding of pollen tubes to a nitrocellulose membrane in the presence of stylar extracts, has been developed (Jauh et al., 1997; Park et al., 2000). This has identified two fractions: one containing a high molecular weight component and the other a small 9 kDa stigma/stylar protein, which was identified as a cysteine-rich adhesin (SCA), that in conjunction with a pectic fraction, enabled pollen adhesion to occur. Cloning and subsequent nucleotide sequence analysis of SCA (Park et al. 2000) revealed that it has homology to plant lipid transfer proteins (Kader, 1997). Immunolocalization confirmed the localization of SCA within the stylar ECM and in pollen tubes growing in vivo, but not grown in vitro (Park et al., 2000). Further studies have resulted in the identification of the other major component (Mollet et al., 2000). Size fractionation indicated that this was a large molecule of approx. 1.5 MDa. Chemical analysis, together with the observation that its pollen adhesion activity was sensitive to treatment with endopolygalacturonase, yet resistant to proteinase K, confirmed it as a member of the pectin family. Since it was recognized by the JIM5 antibody (Knox et al., 1990) it was concluded that it was in part composed of regions of low-esterified homogalacturonan.
Thus, together, the stylar pectin and SCA induce pollen tube adhesion, both to other pollen tubes and to the epidermal cells of the stylar transmitting tract. The bioassay also revealed that the two molecules bind to each other to promote pollen adhesion and stimulate the rate of pollen tube growth. Binding was found to be pH-dependent, with a significant reduction occurring at pH 10, which coincided with a resultant loss in pollen adhesion capacity. These results suggest a role for charge interactions in this phenomenon. However, it appears that this in itself does not fully explain the adhesion activity, as SCA will also bind another negatively charged polymer, alginic acid, but this combination is inactive (Mollet et al., 2000).
These studies lend considerable support to the proposal that a contact-stimulated guidance system facilitates pollen growth through the style (Park et al., 2000). It is suggested that this defines a fixed track down the style, rather than involving any diffusible guidance cue. Lord and co-workers suggest that this mechanism may be considered analogous to the laminin/netrin two-component guidance system that directs the path of neurone outgrowth (Hedgecock & Norris, 1997; Park et al., 2000). In this case, the small, secreted netrin molecule provides an interface between laminin and a receptor in the neurone. If the laminin/neurone analogy were correct then it would be anticipated that there may be specific sites on the pollen with which SCA-pectin complex interacts, possibly involving receptor molecules.
Although there is no direct evidence to support the above hypothesis at present, several receptor molecules have been identified in pollen. A pollen-specific receptor kinase, PRK1, has been identified in Petunia inflata (Mu et al., 1994). However, this has since been implicated in the postmeiotic development of pollen, rather than the interaction of pollen with the extracellular matrix during pollen-tube growth (Lee et al., 1996). Perhaps of more relevance is the identification of a receptor-like kinase LePRK2, isolated from tomato (Muschietti et al., 1998). LePRK2, together with LePRK1, is a serine/threonine kinase found in the pollen plasma membrane/cell wall. Both proteins possess an extracellular domain containing a leucine-rich repeat, a motif that is thought to mediate protein–protein interactions (Kobe & Deisenhofer, 1994). LePRK1 is similar to PRK1 from Petunia (Mu et al., 1994), in that it is expressed during pollen development. LePRK2 is also expressed late in pollen development. However, unlike LePRK1, the protein increases in abundance following pollination, and most interestingly is partially dephosphorylated in a specific response to stylar extracts (Muschietti et al., 1998). The SCA-pectin complex may or may not interact with this class of receptor, nevertheless, these studies indicate that pollen receptors exist that respond to signals emanating from the female tissues. It is anticipated that further pollen receptors will be identified in the future. Finally, the analogy between pollen tube growth and neuronal growth mentioned above has gained additional support recently. Guyon et al. (2000) have produced a study of flavanol-induced pollen germination in Petunia that identified a pollen cDNA encoding a protein whose sequence exhibits similarity to neuromodulin, a neuropeptide implicated in the regulation of axon tip guidance and growth (Zuber et al., 1989). This emerging concept has been reviewed recently (Palanivelu & Preuss, 2000).
In summary, the nature of pollen tube guidance within the style, which was, until recently, very poorly understood, remains a contentious issue. A combination of genetic, molecular genetic and biochemical approaches has led to considerable progress in this area of research. There is certainly very good evidence for complex glycoproteins playing an important role in pollen tube guidance. Furthermore, we have not dealt with the issue of chemical guidance, though there is evidence for this in the literature. Ideas that mechanical, chemical and biochemical gradients may be involved are not mutually exclusive, and it may be that all three play a role. However, it is clear that there are unresolved questions. Part of the problem in this area is the complex nature of the style and the ECM, and the consequent likelihood of further components involved in pollen tube guidance that have yet to be identified. A summary of some of the components identified as involved in growth of pollen through the style is included in Fig. 1.
VI. Control of pollen viability by incompatibility responses
The interactions described above, between pollen and the female sporophytic tissue, have mostly been restricted to the events occurring in compatible pollinations, the aim of which is to ensure fertilization and seed set. The angiosperms have, however, developed means by which to discriminate between desirable pollen and undesirable pollen, which, if allowed to achieve fertilization, would have deleterious effects on the fitness of progeny. These mechanisms have been alluded to in the section describing control of pollen adhesion in the Brassicaceae, where there appears to be discrimination in favour of pollen derived from plants of the same species. In this section we describe progress in our understanding of the mechanisms whereby pollen that is ‘incompatible’ with the recipient plant may be recognized, and subsequently rejected. The first of these mechanisms is interspecific incompatibility, where pollen is rejected from donors of species different to that of the recipient because it is too dissimilar. The second mechanism is intraspecific incompatibility, generally referred to as self-incompatibility (SI) where pollen, although it originates from a donor plant of the same species, is rejected because it is too similar, due to it originating from the same, or a genetically closely related plant. We will deal with the different types of SI first, as it has been much more intensively studied, and in recent years considerable progress has been made in our understanding of this area of control of pollination. However, we will not deal with SI systems that operate at a heteromorphic level, where incompatibility is related to the presence of flowers with differing morphology, such as distyly in Primula spp., since relatively little is known about the molecular basis of incompatibility in these systems. We will then review some aspects of interspecific incompatibility, but it will be apparent that rather less is known about the mechanisms involved in this process in comparison to intraspecific SI.
1. Self incompatibility (SI)
Self incompatibility (SI) is by far the best understood mechanism of pollen recognition and rejection. It is estimated to occur in between 30 and 50% of flowering plant species enabling them to discriminate self-(incompatible) from nonself-(compatible) pollen. Having been distinguished from compatible pollen, self-incompatible pollen is rejected at some point in the pollination process specific to that species. This may happen at hydration, germination, during growth through the style, in the ovule, or even postfertilization in some species. This suggests that there are a variety of mechanisms involved in different SI systems and that the different mechanisms have evolved in isolation. In most cases SI is controlled by a single multiallelic S-locus, which contains genes encoding at least a stylar component and a pollen component. The interaction of these so-called S-gene products determines the fate of pollen growing through the stigmatic and stylar tissues. SI is generally categorized as being under sporophytic or gametophytic control. In sporophytic SI, the pollen S-phenotype is determined by the diploid S-genotype of the parent plant. In gametophytic SI, however, the pollen S-phenotype is specified by its own haploid S-genotype. In some species, there is a more complex genetic control of SI, with the involvement of other loci. For example, SI in the grasses is under control of two loci, S- and Z-. However, these systems will not be dealt with here. In some species, detailed investigations of some of the processes controlling SI have been undertaken. Here we describe the better known SI systems, and what is currently known about their mechanisms.
In the case of single-locus gametophytic self-incompatibility, two mechanistically different systems have been studied in detail at the molecular level. The first of these is the S-RNase system, originally found and extensively characterized in members of the Solanaceae (Anderson et al., 1986; McClure et al., 1989), and more recently reported in the Rosaceae (Sassa et al., 1993), Scrophulariaceae (Xu et al., 1996), and Campanulaceae (Stephenson et al., 2000). A second, different gametophytic SI system is found in the Papaveraceae.
SI in the Solanaceae
In the Solanaceae, in both compatible and incompatible pollinations, the pollen germinates and a pollen tube grows through the transmitting tract of the style. The growth of the incompatible pollen tube is, however, arrested by the time it has reached about one-third of the way through the style. Early studies investigated the stylar proteins of different S genotypes to find proteins associated with SI. Analysis of stylar proteins from Nicotiana alata resulted in the identification of a glycoprotein of c. 30 kDa that exhibited genetic linkage to the S locus (Anderson et al., 1986). Subsequent cloning and analysis of a large number of alleles of this gene from different members of the Solanaceae revealed them to be highly polymorphic, exhibiting between 39 and 98% sequence identity (Anderson et al., 1989; Ai et al., 1990; Clark et al., 1990; Ioerger et al., 1990; Kheyr-Pour et al., 1990; Xu et al., 1990). Some S alleles from different species were found to be more closely related than they were to S alleles from their own species. This finding indicated that SI in the Solanaceae predates the divergence of genera, which in the case of Nicotiana and Petunia is thought to have occurred 27 million years ago. Thus, the S-locus compares with other systems known to be under frequency-dependent selection, such as the MHC loci in vertebrates (Hughes & Nei, 1988), in exhibiting trans-specific polymorphism.
The identity of the pollen S protein in the Solanaceae remains to be determined, but two suggestions as to its nature have been proposed (see Fig. 2). The first model suggests that it is a receptor which internalizes S-RNase molecules in an allele-specific manner (Fig. 2, model I). The S-RNase then degrades pollen rRNA, leading to the cessation of pollen tube growth. In the alternative model (known as the inhibitor model), it is proposed that indiscriminate uptake of S-RNases occurs, and on entering the pollen tube, they interact with the pollen S protein that is proposed to be an RNase inhibitor. This recognizes and inhibits all S-RNases via an interaction with a low-affinity binding site, except when the S-RNase is of the same allelic specificity. In this case, binding occurs via a high affinity, allele-specific site, which somehow prevents the interaction with the low affinity site (Fig. 2, model II). This leaves the S-RNase fully active and able to inhibit pollen tube growth. Although there is no direct evidence to support either hypothesis, accumulating evidence favours the latter, since it can account for the observation that SI breaks down in tetraploid plants (Golz et al., 1999). It is also supported by recent immunocytochemical studies (Luu et al., 2000), which used a mono-specific polyclonal antibody directed against S11-RNase to label pollen tubes growing in incompatible styles. Pollen from a plant homozygous for S12 was used to pollinate styles from a plant of genotype S11S12. Labelling with the anti-S11 antibody detected markedly higher amounts of S11-RNase within the pollen tubes, compared to the extracellular matrix. This suggests that uptake of the S-RNase is independent of S genotype, a finding that adds weight to the inhibitor model.
Another important issue that remains to be addressed is the basis of allelic specificity. Sequence comparisons reveal that S-RNases are organized into five conserved regions and two hypervariable regions that form a continuous surface on one side of the protein. Whether S-specificity resides entirely within the hypervariable regions is debatable. Although the S11 allele of Solanum chacoense was successfully changed into a closely related S13 allele (Matton et al., 1997), hypervariable domain-swap experiments involving more diverged alleles have not proved successful (Verica et al., 1998), but this may be due to the failure of hybrid proteins to fold correctly. This problem is unlikely to be resolved until the tertiary structure of S-RNases is characterized. Work is currently being undertaken towards this goal (Ida et al., 2001; Parry et al., 1998).
Attempts have recently been made to identify other components involved in the SI response in the Solanaceae. McClure et al. (1999) used a differential screening approach to address this problem. A cDNA library was constructed from N. alata stylar tissue. This library was then hybridized with stylar RNA from N. alata and the closely related, but self-compatible species, N. plumbaginifolia. A clone was found that hybridized to N. alata RNA only. The clone designated HT was found to encode a 101 residue asparagine-rich protein. RNA blot analysis suggests that HT transcript accumulates in the style before anthesis with a slight lag behind S-RNase expression. Transgenic experiments demonstrated that HT is essential in the SI response. N. plumbaginifolia × N. alataSc10Sc10 hybrids transformed with an antisense construct of HT lost the ability to reject Sc10-pollen, whereas untransformed control plants were able to reject this pollen. Unfortunately, database searches have failed to detect homology to any known proteins, and, as yet, the function of HT is unknown. The detection of nonS-linked factors involved in the Solanaceae SI response suggests that a true picture of the mechanics of this response is, as yet, incomplete and current searches are underway to identify other elements of the response (McClure et al., 1999).
In conclusion, despite extensive efforts by a number of groups the identity of the pollen S-gene in the Solanaceae and other species with an S-RNase SI system remains to be resolved. Although current evidence now favours the inhibitor model (Fig. 2, model II), a complete understanding of the mechanism of SI in the Solanaceae is still some way off. Whilst isolation of the pollen S-gene remains a major goal, studies by McClure and co-workers have identified other aspects of this system that require further investigation. As mentioned above, their studies have revealed that one or more factors other than the products of the S locus are required for a functional SI system. In addition, they have obtained evidence that reveals a role for S-RNases in interspecific SI (see below).
SI in Papaver
A gametophytic system of SI control is also found in the field poppy, Papaver rhoeas. However, studies at the molecular and biochemical level have revealed that the stigmatic S gene, and the mechanisms involved in pollen inhibition, are completely different from that of the Solanaceae (see Fig. 3). Unlike in the Solanaceae, where the SI response involves the action of cytotoxic proteins, the SI response in P. rhoeas is mediated by a complex signalling cascade. The response does not involve the degradation of rRNA (Franklin-Tong et al., 1991). Instead, several biochemical and cytological changes are triggered in the pollen that are likely to contribute to the inhibition of its growth. The response also has some notable physiological differences to that found in the Solanaceae. P. rhoeas lacks a style and rejection of pollen takes place on the stigmatic surface. As well as this, P. rhoeas has a dry stigmatic surface compared to the wet, lipid-rich exudate found on the surface of the Solanaceae (Elleman et al., 1992). Inhibition of incompatible pollen in P. rhoeas is rapid compared with the relatively slow inhibition in the Solanaceae, acting on a time scale of minutes rather than hours. These fundamental differences have led to suggestions that gametophytic SI systems have evolved independently several times. Research into SI in Papaver has been helped considerably by the development of an in vitro bioassay (Franklin-Tong et al., 1988). The bioassay has permitted the SI response to be reproduced in vitro, and has enabled many downstream events involved in SI-induced pollen inhibition to be analysed in detail. A summary of the P. rhoeas SI response is shown in Fig. 3.
The stigmatic S proteins of P. rhoeas are small (c. 15 kDa) extracellular signalling molecules. In the order of 66 S alleles are estimated to exist in this species (Lane & Lawrence, 1993). Several alleles (S1, S3, S8 and Sn1 from P. nudicaule) of the stigmatic S gene have now been cloned (Foote et al., 1994; Walker et al., 1996; Kurup et al., 1998). Although the exact basis of allelic specificity remains to be elucidated, site-directed mutagenesis of the S1 protein has established that certain residues located in hydrophilic surface loops are crucial for the recognition of S1 pollen (Kakeda et al., 1998; Jordan et al., 1999). These stigmatic S proteins interact with the pollen S gene product, which is believed to be a plasma membrane receptor. The nature of the pollen receptor is unclear at present. One candidate, an S protein binding protein (SBP) has been identified (Hearn et al., 1996). SBP is a pollen plasma membrane glycoprotein of 70–120 kDa. Binding to the S protein is at least partly dependent on the glycan moiety of SBP. Whilst SBP specifically binds S proteins in vitro studies suggest that this binding is not S-allele specific. Hence, SBP may be an accessory receptor rather than the pollen S receptor itself. However, as analysis of S protein mutants has revealed that all mutants that exhibit reduced ability to inhibit incompatible pollen also have reduced SBP binding activity (Jordan et al., 1999), it is conceivable that SBP is the S receptor. Resolution of this issue awaits the cloning of SBP and/or the S receptor.
It is well established that inhibition of incompatible pollen in P. rhoeas is mediated by the activation of a calcium-based signal transduction pathway in the pollen. It is thought that an increase in cytosolic free Ca2+ ([Ca2+]i) is the initial step in the signalling cascade, and that the SI reaction is a receptor-mediated response, with the S protein acting as a signal molecule that triggers increases in [Ca2+]i. Evidence for this comes from calcium imaging studies that demonstrated that increases in [Ca2+]i were stimulated within a few seconds of an incompatible interaction and preceded the inhibition of incompatible pollen tube growth (Franklin-Tong et al., 1993, 1995, 1997). Confirmation that S proteins alone were sufficient to elicit this response came from use of recombinant S proteins, which showed that they acted as signal molecules (Franklin-Tong et al., 1995). As Ca2+ is important for the regulation of pollen tube growth and is known to act as a second messenger in plant cells (Franklin-Tong, 1999a, 1999b; Rudd & Franklin-Tong, 1999; Rudd & Franklin-Tong, 2001), these findings are consistent with expectations.
Downstream of the initial Ca2+ signals, SI-induction results in the phosphorylation of several proteins in a S-specific manner. This strongly suggests that they are involved in SI in Papaver pollen. Two of these proteins have been named p26 and p68 (Rudd et al., 1996, 1997). Phosphorylation of p26, which occurs within 90 s of challenge, is Ca2+-dependent and most likely involves a calcium-dependent protein kinase, CDPK (Rudd et al., 1996). The recent cloning and characterization of p26 indicates that it is a soluble inorganic pyrophosphatase. The pyrophosphatase activity of the p26 recombinant protein is strongly inhibited by phosphorylation. It is proposed that this reduces the rate of pollen tube growth due to an adverse effect on the biosynthetic capacity of the pollen (J. J. Rudd, V. E. Franklin-Tong, F. C. H. Franklin, unpublished). Phosphorylation of p68, in contrast, is Ca2+-independent (Rudd et al., 1997), and provided the first hint that the SI signalling cascade mediating the SI response is complex, having both a Ca2+-dependent and a Ca2+-independent phase. Additional support for this idea comes from recent data that provides evidence that a mitogen activated protein kinase (MAPK) is activated during the SI response in incompatible pollen. The activation of this MAPK appears to be SI-specific and occurs downstream of increases in [Ca2+]i (J. J. Rudd, F. C. H. Franklin, and V. E. Franklin-Tong, unpublished).
A recently identified target for the SI response in P. rhoeas is the actin cytoskeleton. During the response rapid and dramatic changes occur in the actin cytoskeleton of incompatible pollen tubes (Geitmann et al., 2000; Snowman et al., 2000a,b). Within 1–2 min, and possibly within 30 s, there is a reorganization of filamentous (F)-actin to the pollen tube apical region, which is normally relatively free of F-actin bundles. Furthermore, F-actin also accumulates in the cortical region adjacent to the plasma membrane. Concomitant with this is an apparent decrease in the number of F-actin bundles in the lumen of the pollen tube, and by 10 min fine fragments of actin are seen throughout the cytoplasm. Later alterations are seen from c. 20 min up to several hours later in the SI response where pollen F-actin is reorganized into punctate foci, which enlarge over this time period (Geitmann et al., 2000; Snowman et al., 2000a,b). These data strongly suggest that the actin cytoskeleton is a target for the SI signalling cascades. Further downstream of these events, S-specific nuclear DNA fragmentation has been detected 4–12 h after induction of the SI response (Jordan et al., 2000). This is a hallmark feature of programmed cell death (PCD). Nuclear DNA fragmentation is a key target for PCD signalling cascades, and these findings suggest that PCD is triggered by the SI response in Papaver pollen. Thus, the ultimate death of incompatible pollen in this species takes several hours.
A model for how these components and events might participate to elicit SI in Papaver is shown in Fig. 3(a). An indication of the timescale of these events is shown in Fig. 3(b). It is proposed that there are perhaps three phases in the Papaver SI response. First, a very rapid inhibition of tip growth takes place. Since this is known to be reversible, it is thought that a ‘commitment’ phase follows, during which processes are triggered that lead to the irreversible degradative processes detected in the ‘late’ phase.
Broadly speaking, two issues of particular importance require resolution. First, as mentioned above, the nature of the S receptor remains to be determined. Second, it is clear that the mechanism of inhibition of incompatible pollen is highly complex. Substantial progress has been made towards defining aspects of the signal transduction cascade that mediates the SI response, and a number of associated cellular events have been identified. Nevertheless, it still remains to be established how these events, and the signals responsible for them, are interrelated and regulated to bring about the irreversible inhibition of incompatible pollen.
As well as different gametophytic SI systems, it is apparent that there are at least two distinct sporophytic SI systems. One of these is utilized by members of the Convolvulaceae. However, the most intensively studied of the sporophytic systems is that of the Brassicaceae, particularly B. oleracea. Physiologically, the Brassica SI response depends upon the presence of a ‘dry’ stigmatic surface (Dickinson, 1995). Incompatible pollen alighting on the stigma usually fails to hydrate and germinate, the response occurring at the stigmatic surface within minutes of pollination (Ferrari & Wallace, 1975). The growth of any incompatible pollen that succeeds in penetrating the stigmatic surface halts following the deposition of callose produced in the stigmatic papillar cells (Dickinson & Lewis, 1973). These callose deposits associated with the SI response are not, however, required for the response to take place per se (Sulaman et al., 1997). The molecular mechanisms underlying this response are now beginning to be characterized.
SI in Brassica
Molecular studies have revealed that the S locus in Brassica oleracea and B. campestris is complex, extending, in some instances, over several hundred kilobases. Studies of a range of S haplotypes from amongst the 50 or so that are predicted to exist, have identified two stigmatic glycoproteins, a secreted S-locus glycoprotein (SLG) of around 57 kDa (Nasrallah et al., 1985) and a 120-kDa S-receptor kinase (SRK), which contains a region of high homology to SLG (Stein et al., 1991). The exact role of SLG in Brassica SI is unknown, and remains to be ascertained (see below). Meanwhile, it has been demonstrated that SRK is a key element in the Brassica SI system (Nasrallah et al., 1985; Nasrallah et al., 1988; Kandasamy et al., 1989; Stein et al., 1991). It is related to a large family of plant receptor-like kinases (Walker & Zhang, 1990; Stein et al., 1991; Walker, 1993), and direct evidence that the SRK gene encodes a functional serine/threonine protein kinase has been obtained (Goring & Rothstein, 1992).
Nucleotide sequencing reveals that SRK genes can be classified into two groups that correlate with the phenotypic dominance/recessive relationships in plants carrying various S allele combinations. Class I alleles exhibit a strong SI phenotype and are generally dominant or codominant, whereas Class II alleles are usually recessive and exhibit a weaker SI phenotype. Alleles within each group exhibit c. 90% homology, and approx. 70% homology with members of the other group. Initially, sequencing indicated that for any given S-haplotype, the extracellular domain of SRK was highly similar to the SLG encoded by the same S -haplotype. However, more recent studies indicate that in some S-haplotypes there is significant sequence divergence. Initially, it was generally believed that SRK and SLG worked in conjunction in the SI reaction, possibly as a receptor complex. The analysis of self-compatible mutant lines has confirmed that a functional SRK is required for rejection of incompatible pollen (Goring et al., 1993; Nasrallah et al., 1994; Takasaki et al., 2000).
Most S-haplotypes appear to encode a functional SLG gene, and in most cases it appears to be expressed. Recent studies have attempted to ascertain the function of SLG. Dixit et al. (2000) demonstrated that expression of SLG alongside SRK was necessary for the accumulation of physiologically significant quantities of SRK in transgenic tobacco plants. Without coexpression of SLG and SRK, SRK formed high molecular weight aggregates. This implies a role for SLG in maintaining the solubility and stability of SRK. Transgenic studies by Takasaki et al. (2000) also showed a potential role for SLG in the enhancement of the SI response, although SLG alone was unable to confer S-specificity. However, in certain S-haplotypes SLG activity may be dispensable, as a number of SI lines have been identified that lack detectable SLG protein (Roberts et al., 1994; Gaude et al., 1995).
A further member of the S gene family, SLR1 (S-locus related), has also been identified. However, the SLR1 gene does not map to the S locus, and is therefore, not directly involved in SI. Instead, as mentioned earlier, it appears to have a role in the adhesion of pollen to the stigma (Lalonde et al., 1989; Trick & Flavell, 1989; Boyes et al., 1991).
Since the identification of pistil S genes from Brassica spp., all of the major groups investigating SI in Brassica have been searching for the male counterpart, the pollen S gene. Until recently, very little was known regarding the male side of the interaction. As functional SI requires that both male and female components exhibit complete genetic linkage, it was considered likely that they were in close physical proximity to one another. Having isolated part of the S locus containing SRK, it was then theoretically possible to identify potential pollen S candidates by sequencing regions around SRK. Nevertheless, the physical size and complex nature of the S locus has meant that this route was employed only recently, following the delineation of the S locus by Casselman et al. (2000). Subsequently, a gene, SCR (S locus cysteine rich protein), mapping to the S locus has been cloned and shown in transgenic plants to confer pollen S specificity (Schopfer et al., 1999). The identification of SCR was achieved during sequencing of the 13 kb region between the S8 SRK and SLG genes. The SCR is secreted as a mature hydrophilic protein of 8.4–8.6 kDa, and is characterized by the presence of 8 cysteine residues (Schopfer et al., 1999).
Initial identification of SCR as a candidate for the pollen S gene was undoubtedly aided by it possessing a similar structure to the PCPs. The PCPs, although not S-linked, had previously been shown to interact with SLG (see Section II). Just before the publication of the SCR sequence, Suzuki et al. (1999) reported the analysis of a B. campestrisS-haplotype, and commented that one of the ORFs, SP11 (here 11 refers to the size of the polypeptide and not the S-haplotype), also had similarities to the PCPs previously identified. Comparison of this gene with the SCR allele suggests that SP11 is, in fact, the pollen SI determinant in B. campestrisS9. Indeed, this has now been confirmed by studies in which application of recombinant SP11 from the S9 haplotype of B. campestris has been shown to elicit the SI response on S9 stigmas in the presence of cross pollen (Takayama et al., 2000).
A model for SI in Brassica has been proposed (see Fig. 4), although it is clear that several components remain to be identified. It is proposed that interaction of SCR and SRK triggers a signal transduction cascade in the stigmatic papillar cell, which results in rapid inhibition of pollen growth (as shown in Fig. 4). Several stigmatic cellular targets have been identified, using yeast two-hybrid screening to find proteins that interact with the kinase domain of SRK (Bower et al., 1996; Gu et al., 1998). One of these, ARC1, interacts with SRK in a phosphorylation-dependent manner (Gu et al., 1998). Another is THL1, which has been reported to regulate autophosphorylation of SRK (Cabrillac et al., 2001). Since down-regulation of ARC1 in transgenic plants results in the partial breakdown of SI, this suggests that it plays an important role (Stone et al., 1999). However, knocking-out ARC1 expression does not result in full self-compatibility, as the level of seed set is not equivalent to that obtained from a cross-pollination. This suggests that other unidentified components participate in the rejection mechanism. The identification of a self-compatible line of Brassica that is defective in an aquaporin–like stigmatic gene, mod, suggests the SI response may involve the regulation of water transfer from the stigmatic papillae to the pollen (Ikeda et al., 1997). This component is not linked to the S-locus, so it probably acts as a modifier gene. However, since it has been shown to be required for SI to take place, it clearly plays an important role in the response, most likely very early, at the pollen hydration stage.
SI in Ipomoea
Investigations are also currently underway to determine the molecular nature of sporophytic SI in the Convolvulaceae using Ipomoea trifida, a close relative of sweet potato. As the SI system in the Convolvulaceae has a genetical control that is identical to the Brassicaceae (Kowyama et al., 1980) it might be expected to employ similar molecules in this cell–cell recognition process. It appears, however, that Ipomoea does not employ the same genes and mechanisms as Brassica to control SI. Although SRK homologues have been identified in I. trifida, none of these has been demonstrated, thus far, to be S-linked. It is thought, for this reason that the underlying mechanisms of these two families are unrelated (Kowyama et al., 1996), and that SI in the Brassicaceae and the Convolvulaceae probably evolved independently.
Attempts to identify the molecular basis of SI in this family have concentrated on the use of two strategies. First, 2D-SDS-PAGE gel analysis of proteins derived from the stigmatic tissues of plants of different S-genotypes identified polymorphic protein spots of c. 70 kDa, designated as SSPs (S-locus-linked stigma proteins). Full-length cDNA clones corresponding to these polymorphic proteins were then isolated. Homology searches with the sequence of the four SSP genes identified them as having structural homology to nonmetallo short-chain alcohol dehydrogenases (SCADs). However, linkage analysis later showed recombination between SSP and the S-locus. Second, AFLP-fingerprinting has been attempted, using stigma cDNA derived from plants of several different S-genotypes. From these studies 11 putative S-linked fragments were identified, one of which was found to be SSP. RFLP analysis is currently being carried out on the remaining AFLP clones, although, as yet, no candidate S-genes have been documented and thus the molecular basis of SI in the Convolvulaceae remains unknown (Kowyama et al., 2000).
2. Interspecific incompatibility responses
By contrast to intraspecific incompatibility (SI), interspecific incompatibility has received far less attention. Hence, comparatively little is known regarding the molecular mechanism that underlies this process. Two general models have emerged to describe how angiosperms distinguish con-specific pollen from that produced by donors of foreign species. The incongruity model, proposed by Hogenboom (1975), suggests that interspecific incompatibility is the result of barriers determined by evolutionary divergence of physiology or morphology between species. This is especially likely in those species that are only distantly related. Incongruity then, can be seen as an essentially passive process. The alternative model contrasts with incongruity, since it is proposed that cross-hybridization is prevented by an active process that inhibits what would otherwise be a compatible pollination (de Nettancourt, 1977). These interspecific incompatibility responses have been observed within several plant families.
A connection between the underlying mechanisms behind interspecific incompatibility and SI has been suggested (Lewis & Crowe, 1958). Within many plant families, there are closely related species exhibiting both SI and self-compatibility (SC). An important feature of interspecific crosses in a number of genera is that they allow for crosses in one direction only, with the reciprocal cross being unsuccessful (Mutschler & Liedl, 1994). For instance, it has frequently been observed that pollen from self-compatible species is often rejected by pistils of self-incompatible species, whilst the reciprocal crosses are viable. This is generally referred to as the SI × SC rule. However, this apparent interrelationship between interspecific incompatibility and SI is not clear-cut. There are exceptions to the SI × SC rule and the timing and nature of the response in interspecific incompatibility is not necessarily equivalent to that in the SI response (Ascher & Peloquin, 1968). Nevertheless, there is evidence suggesting that factors involved in interspecific incompatibility are linked to the S locus. For example, Pandey (1981) demonstrated differences in ability to reject foreign pollen between S alleles of the same Nicotiana species, whilst mapping studies have confirmed a role for interspecific incompatibility for the S locus in Lycopersicon (Chetelat & de Verna, 1991). Recent evidence points to a more complex picture regarding interspecific incompatibility. For example, a cross between N. alata and the closely related species, N. tabacum shows an exception to the SI × SC rule in that although this cross is rejected, the same cross using a self-compatible cultivar of N. alata is also nonviable. Transgenic studies, examining the role of S-RNases in interspecific incompatibility, indicate a requirement for both S-RNases and additional factors that remain to be identified (Murfett et al., 1996). Thus, even in systems where there is substantial evidence for a link between SI and interspecific incompatibility, the S-locus describes only a part of the system (McClure et al., 2000). The challenge facing workers in this field is to establish the nature of those other factors and to determine the extent of the involvement of the S-locus in interspecific incompatibility.
VII. Conclusions and perspective
We have attempted to review the considerable progress made over the past decade in elucidating the components and molecular processes involved in pollination events. What is apparent is that this involves highly complex cellular interactions that regulate a series of crucial events, such as pollen adhesion, hydration, pollen tube growth and guidance to the ovules before fertilization may be achieved (as summarized in Fig. 1). Over and above this, there are incompatibility mechanisms (see Figs 2, 3 and 4) that provide barriers to fertilization in many instances. This review shows that in many areas there are just a few studies, which although very informative, do not give a very coherent picture overall. Our knowledge in some aspects is therefore rather fragmented. Furthermore, the complexity of both the intercellular and intracellular interactions is only just beginning to be revealed. Although it is apparent that much progress has been made in recent years in our understanding of pollen–pistil interactions, there still remains huge gaps in our knowledge.
Work in the authors’ labs is funded by the BBSRC and the Gatsby Technical Education Project.