|Home | About | Journals | Submit | Contact Us | Français|
In contrast to animals and lower plants such as mosses and ferns, sperm cells of flowering plants (angiosperms) are immobile and require transportation to the female gametes via the vegetative pollen tube cell to achieve double fertilization. The path of the pollen tube towards the female gametophyte (embryo sac) has been intensively studied in many intra- and interspecific crossing experiments with the aim of increasing the gene pool of crop plants for greater yield, improved biotic and abiotic stress resistance, and for introducing new agronomic traits. Many attempts to hybridize different species or genotypes failed due to the difficulty for the pollen tubes in reaching the female gametophyte. Detailed studies showed that these processes are controlled by various self-incompatible (intraspecific) and cross-incompatible (interspecific) hybridization mechanisms.
Understanding the molecular mechanisms of crossing barriers is therefore of great interest in plant reproduction, evolution and breeding research. In particular, pre-zygotic hybridization barriers related to pollen tube germination, growth, guidance and sperm delivery, which are considered the major hybridization controls in nature and thus also contribute to species isolation and speciation, have been intensively investigated. Despite this general interest, surprisingly little is known about these processes in the most important agronomic plant family, the Gramineae, Poaceae or grasses. Small polymorphic proteins and their receptors, degradation of sterility locus proteins and general compounds such as calcium, γ-aminobutyric acid or nitric oxide have been shown to be involved in progamic pollen germination, adhesion, tube growth and guidance, as well as sperm release. Most advances have been made in the Brassicaceae, Papaveraceae, Linderniaceae and Solanaceae families including their well-understood self-incompatibility (SI) systems. Grass species evolved similar mechanisms to control the penetration and growth of self-pollen to promote intraspecific outcrossing and to prevent fertilization by alien sperm cells. However, in the Poaceae, the underlying molecular mechanisms are still largely unknown.
We propose to develop maize (Zea mays) as a model to investigate the above-described processes to understand the associated intra- and interspecific crossing barriers in grasses. Many genetic, cellular and biotechnological tools including the completion of a reference genome (inbred line B73) have been established in the last decade and many more maize inbred genomes are expected to be available soon. Moreover, a cellular marker line database as well as large transposon insertion collections and improved Agrobacterium transformation protocols are now available. Additionally, the processes described above are well studied at the morphological level and a number of mutants have been described already, awaiting disclosure of the relevant genes. The identification of the first key players in pollen tube growth, guidance and burst show maize to be an excellent grass model to investigate these processes in more detail. Here we provide an overview of our current understanding of these processes in Poaceae with a focus on maize, and also include relevant discoveries in eudicot model species.
The initial interactions between the pollen grain (male gametophyte) and the sporophytic receptive papilla cells of the feathery grass stigma (Fig. 1) appear rather unspecific, but generally occur very quickly. The time between pollen capture and germination is <5 min in many grass species analysed (Heslop-Harrison, 1979). Capture of airborne grass pollen starts with the interaction between the male gametophyte and the sporophyte (Fig. 2). Although the structural components of the pollen grain and the stigma surface are well described in many grass species including maize (Heslop-Harrison, 1982), the detailed physiological and molecular mechanisms of pollen adhesion remained unclear. Stigma proteins appear to play a crucial role in this process because adhesion in maize is strongly reduced once stigma surface proteins are removed or masked (Heslop-Harrison and Heslop-Harrison, 1981). However, even pollen grains of incompatible genotypes of maize or other grass species such as fragile oat (Gaudinia fragilis) and rye (Secale cereale) germinate normally before tubes are arrested at or near the stigma surface (Heslop-Harrison and Heslop-Harrison, 1982), indicating that these proteins are not involved in incompatibility responses. Moreover, pollen of unrelated plant species such as Arabidopsis thaliana are capable of adhering to the grass stigma (although interactions are less strong; Fig. 2E) and germinate between 45 and 60 min after adhesion. Remarkably, they germinate at a high frequency and grow on the surface of maize stigmatic papilla cells (Lausser et al., 2010), indicating that species-specific cross-talk between pollen grains and stigmatic papilla cells plays a minor role in grasses.
Since pollen grains shrink and desiccate to a water content of usually only 10–30 % during maturation (Heslop-Harrison, 1979), they need to take up water from the stigma papilla at relatively high rates. Grass pollen in particular rehydrates very quickly within only 1–3 min. Hydration and gelation of the pectocellulosic zwischenkörper is a prerequisite to open the grass pollen grain. The pollen tube tip then emerges from the single grass pollen aperture and gives rise to the pollen tube generated by the vegetative tube cell, which engulfs and delivers twin sperm cells towards the female gametophyte. Maize pollen grains, like those from other grass species, usually germinate within 5 min. Transfer of the sperm cells from the pollen grain into the tube is delayed and does not occur until 40–60 min after germination in vitro, when pollen tubes are already 300–500 µm in length (Kliwer and Dresselhaus, 2010). It is noteworthy that the above-mentioned early events of the pollination process do not necessarily require the presence of a stigma. Pollen from all commonly used model species, dicot and monocot, can germinate in vitro if the concentrations of sugar, boron, calcium and protons and the humidity for germination on plates are chosen properly for every species. In conclusion, pollen capture, adhesion, hydration and germination do not seem to represent a pre-zygotic hybridization barrier in maize and other grasses.
Although a number of pollen grains can be supported by the stigma to hydrate and germinate, usually only one pollen tube penetrates the micropyle to perform fertilization in maize (Heslop-Harrison, 1982; Lausser et al., 2010). Thus, after a heavy pollination, the numbers achieving passage must be guided precisely, but additionally dramatically reduced at various points during the pollen tube pathway. The latter happens in maize (1) by competition on the receptive trichomes and (2) inside the transmitting tracts; (3) by elimination of the late-entering tubes at the stigma abscission zone; and (4) at a constricted zone of the transmitting tract in the upper ovary wall (Heslop-Harrison et al., 1984). As a result, pollen tubes are displaced onto the stigma surface or blocked in the stylodia or upper ovary wall, where they stop growing and eventually fuse.
Maize silks, which can be regarded as stigmata, are extremely long compared with those of eudicot species, but similar to the twin-branched and feathery stigmata of other grasses (Fig. 1). The major difference is that the maize stigma is particularly extended and fused over most of its length, thus containing two transmitting tracts, one in each silk half (Kiesselbach, 1999; Lausser et al., 2010). Stigma invasion of pollen tubes begins with the crossing of the cuticle of the stigma and invasion of the intercellular space of the sporophytic tissues (Heslop-Harrison, 1982). It has been shown that xylanase XYN1, hydrolases and class B β-expansins are required for stigma invasion in maize (Suen and Huang, 2007; Bedinger and Fowler, 2009; Valdivia et al., 2009). Interspecific cross-pollination of maize silks or those of its close relative eastern gamagrass (Tripsacum dactyloides) with pollen from A. thaliana or other unrelated species showed that pollen germinated readily but failed to invade the stigma of maize and Tripsacum (Lausser et al., 2010). So far, the picture of stigma invasion by the pollen tube remains incomplete in maize as well as in other Poaceae. The higher specificity of this process compared with the pollination processes described above reveals invasion as the first gatekeeper in terms of hybridization control and pathogen defence.
The stigma surface is the first but not the only obstacle to a pollen tube on its pathway towards the female gametophyte. The silk of maize, which is up to 20 cm in length, offers growth support for the pollen tube since pollen carries resources for a maximum tube length of only about 2 cm (Heslop-Harrison et al., 1984). Furthermore pollen tubes have to target the transmitting tracts inside the sporophytic silk tissues. Unlike the anatomy of the stigma and style of eudicot species, where pollen tubes directly enter the transmitting tissue after invasion of the stigma, grass pollen tubes enter the transmitting tract after invasion by crossing the stigma cortex in a stepwise traverse across successive cell files (Fig. 2A; Heslop-Harrison, 1982). Sporophytic pollen tube guidance is involved in this process and will be discussed below. In summary, pollen tube guidance towards the transmitting tract seems to represent a second pre-zygotic hybridization barrier.
The third hybridization barrier appears to be pollen tube growth support inside the transmitting tract in maize. While growth of Tripsacum pollen tubes outside the maize transmitting tract is arrested after about 2 cm growth, probably after pollen resources are consumed, Tripsacum tubes inside the tracts grow significantly longer before arrest (Lausser et al., 2010). Similarly, incompatible ga1 (gametophytic factor1) maize pollen tubes target Ga1-s/Ga1-s transmitting tracts, but growth ceases after a few centimetres. Details are discussed in relation to genetic cross-incompatibility (CI) in grasses later in this review. The use of additional maize mutants such as white pollen1 (whp1) and colorless2 (c2) indicated that flavonols are also required for pollen growth in the style (Pollak et al., 1995). Both genes encode chalcone synthase, the key enzyme of the flavonol biosynthetic pathway. Maize pollen tubes deficient in flavonols display normal germination, but retarded growth in the stigma. This is a unilateral defect that cannot be rescued by wild-type silks, but can be rescued by exogenously applied flavonol aglycones at pollination (Pollak et al., 1995). Considering that C2/C2 silks do not contain flavonols, it is likely that pollen tube-generated flavonols are required to stimulate growth support and nutrient/sugar supply by the sporophytic tissues of the stigmata.
We occasionally found disoriented pollen tubes, which reached the transmitting tracts but grew towards the silk tip and continued growth, indicating that there are no guidance cues once pollen tubes grow inside the tract. This suggests that growth inside the tract is mainly under mechanical control of the small and elongated fusiformous tract cells (Fig. 2B). Competition among pollen tubes inside the transmitting tracts seems less significant because >10 tubes were visible inside maize tracts (Fig. 2C). After heavy pollination, a similar number also appeared on the surface of the inner integument inside the ovary after leaving the tracts (Lausser et al., 2010). In eudicot model plants, such as wishbone flower (Torenia fournieri) and A. thaliana, pollen tubes become competent for female gametophyte signals during transmitting tract growth (Higashiyama et al., 2006; Palanivelu and Preuss, 2006). Molecular studies supported these findings as the gene expression pattern of pollen tubes grown in vitro differ from that of pollen tubes grown through the stigma and style (Qin et al., 2009). In contrast to these model species, in vitro grown maize pollen tubes are attracted by ovules at high rates (M. L. Márton and T. Dresselhaus, unpubl. res.), indicating that stylar growth is not a prerequisite for maize and possibly other grasses to become competent to respond to female signals.
Flowering plants have developed a number of self-incompatibility (SI) and cross-incompatibility (CI) pollination mechanisms to prevent fertilization by self or alien pollen tubes, respectively. SI promotes intraspecific outcrossing to increase the gene pool and to avoid the negative effects of genetic homozygosity. Seed and F1 hybrid sterility, which is usually accompanied by wide crosses (Sharma, 1995), is avoided in CI reactions to ensure fertilization of compatible gametes and to generate a maximum number of offspring. Both mechanisms probably played an important role in the evolution, reproductive isolation and thus diversification of plants including the grasses. While the molecular mechanisms involved in CI are not well understood, SI in most plant families investigated relies on a single polymorphic S- or sterility locus (Rea and Nasrallah, 2008).
The best studied SI systems are the sprorophytic self-incompatibilty (SSI) system of the Brassicaceae and the gametophytic self-incompatibility (GSI) systems of the Solanaceae, Rosaceae, Plantaginaceae and Papaveraceae. In the Brassicaceae the S-locus encodes the polymorphic pistil S-determinant SRK (S-locus receptor kinase) and a highly polymorphic small cysteine-rich protein ligand SCR/SP11 (S-locus cysteine-rich protein) representing the pollen S-determinant (Chapman and Goring, 2010). Solanaceae, Rosaceae and Plantaginaceae rely on an S-RNase GSI system named after the pistil S-determinant encoding a secreted RNase, probably acting as a cytotoxin. The pollen S-determinants SLF/SFB encode F-box proteins, known for their role in ubiquitin-mediated protein degradation (Kumar and McClure, 2010). In compatible reactions it is thought that S-RNases are quickly degraded after uptake in pollen tubes, involving the activity of the S-locus-encoded F-box proteins. Finally, in the Papaveraceae, the female S-determinant PrsS (Papaver rhoeas stigma S-determinant) was identified as a small polymorphic protein secreted by the stigmatic papilla cells, and the male determinant PrpS (P. rhoeas pollen S-determinant) as a small pollen tube-located transmembrane protein probably regulating Ca2+ influx and pollen tube cell death via activation of a Ca2+-dependent signalling cascade (Wheeler et al., 2010).
Surprisingly little is known about the molecular mechanisms of SI in the Poaceae. SI is widely distributed in grasses and has been reported to occur in at least five of the 13 subfamilies of the Poaceae. It is very common among forage grasses (Connor, 1979; Baumann et al., 2000; Yang et al., 2008). With the exception of rye (Secale cereale) SI has not been reported in any major grass crops including maize, rice, wheat, barley and oat, but it occurs in their wild relatives. Lundquist (1954) and Hayman (1956) were the first who concluded from genetic experiments with rye and blue canary grass (Phalaris caerulescens) that SI in grasses is gametophytic (GSI) and regulated by two polyallelic loci, S and Z. Subsequent studies with other grass genera indicated that this system is transmittable to the whole family (Heslop-Harrison, 1982). As a consequence of a two-locus system, SI in grasses has features distinct from single-locus systems which include (1) maintenance of SI at the polyploidy level; (2) differences in the penetrance of compatibility between two individuals and (3) between reciprocal crosses similar to unilateral cross-incompatibility (UCI) described below; and (4) the occurrence of homozygosity at one of the two loci (Yang et al., 2008). Although none of the gene products of the two loci S and Z have been identified to date, high-resolution mapping and map-based cloning approaches combined with comparative genetics recently resulted in the identification of a number of potential male and female SI candidate genes of the S–Z SI system in perennial ryegrass (Lolium perenne) (Yang et al., 2009; Shinozuka et al., 2010), wild bulbous barley (Hordeum bulbosum) (Kakeda et al., 2008), rye (Hackauf and Wehling, 2005) and blue canary grass (Bian et al., 2004). Interestingly, in rye, a homologue of Arabidopsis AtUBP22, a ubiquitin-specific protease, specifically expressed in pistils and co-segregating with the Z-locus, was identified as a candidate SI gene (Hackauf and Wehling, 2005). Considering that ubiquitination and protein degradation play a major role during SSI and GSI in eudicot species, this gene appears promising for more detailed investigations. However, functional confirmation for all candidate grass GSI genes still lacks experimental proof.
Grass pollen germinates very quickly and SI reactions of some incompatible genotypes of species such as fragile oat, blue canary grass and rye are already arrested at or near the stigma surface often within 2 min after germination (Heslop-Harrison, 1979; Shivanna et al., 1982). This phenomenon is similar to SSI in the Brassicaceae. However, pollen tube growth arrest in grasses is determined by their own genotype and not by the maternal sporophytic tissues of the anther. This indicates that the SI mechanism is different, but may include similar molecular players such as polymorphic pollen tube secreted protein ligands and stigmatic receptor-like kinases (RLKs) (Wehling et al., 1994; Yang et al., 2008).
More commonly, pollen tubes of SI species such as wild bulbous barley penetrate the stigma surface and rejection occurs inside the stigmata or even after arriving at the transmitting tract in species such as meadow foxtail (Alopecurus pratensis) (Heslop-Harrison, 1982; Shivanna et al., 1982). These primary interspecific SI barriers have been associated with GSI, although they overlap in appearance with primary CI barriers reported for maize and other species. Late-acting CI barriers in most intergeneric crosses between grasses fail to form hybrid zygotes, and pollen tubes are usually arrested in the proximity of or inside the transmitting tract (e.g. Heslop-Harrison, 1982; Lausser et al., 2010). Most studies on late-acting CI have been conducted in maize and resulted in the identification of several genetic loci leading to CI: almost all North American flint and dent maize varieties, for example, are genetically ga1/ga1 and generate pollen that is not able to fertilize many popcorn and Central American maize races which are Ga1-s/Ga1-s (Nelson, 1952). The reciprocal cross is fully fertile. Such a UCI system was first described by Demerec (1929) in maize. The Ga1 system is to date the best investigated system: House and Nelson (1958) were the first to show that 32P-labelled ga1 pollen tubes germinated normally and grew into the stigmata of homozygous Ga1-s plants, but growth was progressively slower and finally ceased. These experiments were recently extended (Lausser et al., 2010) and it was shown that the majority of pollen tubes targeted the transmitting tract, but pollen tube growth support did not occur, leading to arrest about 4–6 cm from the germination site. In the reciprocal cross, growth of Ga1-s pollen was not affected on ga1/ga1 stigmata. Two further UCI systems have been described for the Teosinte crossing barrier1 (Tcb1) and the Gametophytic factor2 (Ga2) loci that restrict crossability of maize (Zea mays ssp. mays) pollen to silks of some annual maize subspecies teosinte (Z. mays ssp. mexicana) populations (Evans and Kermicle, 2001; Kermicle and Evans, 2010). The reciprocal crosses generate highly fertile offspring. Genetic studies have shown that the pollen–pistil barrier regulated by Tcb1 resulted from incongruity rather than active rejection as the pistil requires the presence of a matching allele in pollen (congruity) for successful fertilization (Kermicle and Evans, 2005). Although the precise crossing barrier is unknown, genetic studies demonstrated that Tcb1 and Ga1-s are genetically linked but different loci. It was further shown that cross-recognition seems to occur between the two systems and that the effect is pre-zygotic (Evans and Kermicle, 2001). Ga2-s is another genetic locus acting pre-zygotically and was suggested to encode a major player to protect teosinte ovules against maize pollen (Kermicle and Evans, 2010). Taken together, Ga1-s, Ga2-s and Tcb1-s represent candidate speciation genes regulating reproductive isolation of teosinte ovules from maize pollen. An even more complex UCI system that is controlled by three recessive loci has been reported to act in maize (Rashid and Peterson, 1992). Of these Cif (cross-incompatibility female) controls the UCI reaction in the female parent, while the other two loci Cim1 and Cim2 (cross-incompatibility male) regulate the reaction in the pollen parent. It is noteworthy that this cross is only incompatible if the female parent is cif/cif and the pollen parent homozygously recessive for cim1 and cim2. UCI was also observed after backcrossing between wild (Oryza rufipogon) and cultivated (O. sativa) rice strains (Matsubara et al., 2003). Although the nomenclature of the maize Cif/Cim UCI system was applied, cytological investigations showed that UCI was post-zygotic in rice. Growth of pollen tubes of eastern gamagrass (T. dactyloides), a close relative of maize that does not naturally cross with maize, showed growth retardation and arrest in the cortex of the stigmata and within transmitting tracts, the latter reminiscent of growth arrest of the UCI Ga1 system (Lausser et al., 2010). Similar observations have been made in crosses between millet (Pennisetum glaucum) and sorghum (Sorghum bicolor) (Heslop-Harrison, 1982), supporting the hypothesis that growth support requires communication between pollen tubes and the sporophytic transmitting tract cells. CI between bread wheat and wild barley was shown to be controlled by three Kr genes, leading to late pollen tube growth arrest at the base of the stigma (Snape et al., 1980).
In conclusion, maize is a good model to understand CI and/or selective fertilization as well as reproductive isolation in grasses and flowering plants in general. It will be important to elucidate whether primary and late CI as the first mechanisms are similar to GSI in grasses and may involve similar molecular players. Whether the UCI mechanism described above has played a role in the evolution and reproductive isolation of Zea species and to what extent it applies to other grasses also remains to be investigated.
As described above, pollen tubes of grasses seem to be targeted to the transmitting tract by signalling molecules in a process which is commonly referred to as chemotaxis. In contrast to eudicot pollen tubes, which have to target one of many unfertilized ovules located within the ovary, a grass ovary encloses only one single bitegmic ovule, and pollen tubes are delivered to the ovarial cavity directly by following the path of the transmitting tract towards the single ovule. The lack of a funiculus in the uniovulate grass ovary questions the necessity for and range of ovule-derived signalling molecules to attract and support growth of grass pollen tubes.
Unlike in eudicots where pollen tubes are directly targeted into the transmitting tract via the anatomy of the stigma, species-specific chemotactic factors were already postulated >35 years ago to be required in grasses (Heslop-Harrison et al., 1985). It was suggested that these molecules are secreted by the transmitting tract cells, to guide self, but not alien pollen tubes in maize and other grass species. The finding that a large percentage of Tripsacum pollen tubes are mistargeted to the stigma cortex instead of reaching the transmitting tract of maize silks, whereas in the reciprocal cross maize pollen tubes occur exclusively in the transmitting tract of Tripsacum (Lausser et al., 2010), provides support for the existence of species-specific pollen tube attractants secreted by the transmitting tract of grasses. Similar findings were observed when maize stigmata were pollinated with rye pollen, resulting in germination and stigma invasion, but rye pollen tubes were unable to enter the maize transmitting tracts (Heslop-Harrison, 1982). This indicates that this particular pollen tube growth phase represents another important hybridization barrier in grasses for compatible or incompatible pollinations. Until now the molecular identity of the communication molecules involved remained elusive (Lausser and Dresselhaus, 2010). Only two peptides have been identified in lily, chemocyanin (an alkaline plantacyanin protein generated most abundantly in the stigma and style) and a stigma/stylar cysteine-rich adhesin (SCA; a plant lipid transfer protein), which appear to act in concert as chemotropic attractants directing growing pollen tubes towards and into the hollow style (Kim et al., 2003; Park and Lord, 2003). While chemocyanin has been shown to attract pollen tubes in vitro (Kim et al., 2003), thus acting as a chemotactic guidance cue, SCA by itself does not display chemotactic effects. SCA seems to function as an accessory protein, a hapotactic guidance cue, because chemocyanin attraction activity increased when mixed with SCA (Park et al., 2000; Kim et al., 2003). The single PLANTACYANIN gene in Arabidopsis shows a high similarity to lily chemocyanin (86·8 % at the amino acid level), and overexpression caused aberrant growth of wild-type pollen tubes on the stigma in about 50 % of cases (Dong et al., 2005). However, its ability to attract pollen tubes in vitro remains to be demonstrated. Several predicted chemocyanin-like proteins exist in maize (Alexandrov et al., 2009) and it would be interesting to find out whether they also play a role in sporophytic pollen tube guidance or if they are involved in other processes. Transmitting tissue-specific glycosylated arabinogalactan proteins TT (transmitting tract proteins) found in tobacco styles have also been shown to be involved in pollen tube guidance (Cheung et al., 1995; Wu et al., 1995). However, it is still unclear whether these proteins control pollen tube directional growth or merely sustain its elongation (Higashiyama and Hamamura, 2008).
In Arabidopsis, the final phases of pollen tube guidance are defined as pollen tube exit from the transmitting tract and growth on the septum surface, funiculus and micropylar region of the ovule (Johnson and Preuss, 2002). These phases depend on the presence of a functional ovule and involve both sporophytic and gametophytic attraction signals. GABA (γ-aminobutyric acid) was proposed as a candidate for the sporophytic guidance cue, because it forms a gradient in the pistil, with the highest concentration at the inner integument of the Arabidopsis ovule (Palanivelu et al., 2003). Pollen tubes were more sensitive to higher concentrations of GABA and guidance was impaired after self-pollination in the pollen–pistil interaction2 (pop2) Arabidopsis mutant. The POP2 gene encodes a transaminase that degrades GABA (Palanivelu et al., 2003). However, until now, a role for GABA consistent with these observations has not been demonstrated, and pollen tube guidance in a pistil with low GABA levels has not been shown.
In grasses in the absence of the female gametophyte no clear bias in pollen tube growth behaviour could be observed. Since a funiculus is lacking in maize, it was postulated that pollen tube growth after transmitting tract exit towards a distinct area at the micropyle is exclusively under sporophytic control (Lausser et al., 2010). Furthermore, due to the anatomical architecture of the grass ovule surface, i.e. the orientation of integument and nucellar cells towards the nucellar cone of the micropyle, the direction of pollen tube growth in this phase may rely on mechanical cues as there is no evidence for chemotaxis in this phase. Therefore, it appears likely that female gametophyte-derived chemotactic long-range factors either do not exist or play a minor role in maize and possibly other grasses.
The last step of the pollen tube journey, namely micropylar pollen tube guidance, and downstream events, including sperm cell release, are controlled by the female gametophyte. This phase involves numerous inter-gametophytic communication events (for reviews, see Dresselhaus, 2006; Dresselhaus and Márton, 2009). In female gametophyte mutants of Arabidopsis with less severe or late developmental defects, pollen tubes grew along the funiculus, but did not enter the micropyle (Shimizu and Okada, 2000). Recently, in maize mutants lacking embryo sacs, pollen tubes were observed arriving in the close vicinity of the nucellar cone of the micropyle, but, instead of penetrating, they continued growth in random directions (Lausser et al., 2010). These results, in addition to those obtained from the in vitro Torenia system (Higashiyama et al., 1998), suggest that female gametophyte-controlled pollen tube guidance in eudicots is regulated by a two-step process involving funicular guidance (guidance from the surface of the placenta to the funiculus) and micropylar guidance (guidance from the entrance of the micropyle to the female gametophyte) (Shimizu and Okada, 2000; Higashiyama et al., 2003; Higashiyama and Hamamura, 2008). In grasses a funiculus is lacking. Therefore, pollen tubes merely need to be guided to the micropyle and further towards the egg apparatus. As a consequence gametophytic pollen tube attraction in Poaceae only involves the latter micropylar guidance step (Lausser and Dresselhaus, 2010). In addition, in Arabidopsis and Torenia, pollen tubes easily reach the micropylar region of the female gametophyte consisting of the egg cell and two synergids, or the naked embryo sac that protrudes through the micropylar opening of the integuments, respectively. In contrast, in the Poaceae the female gametophyte is additionally embedded in a few layers of stack-like nucellus cells, representing an additional obstacle before pollen tubes can finally enter the embryo sac and execute double fertilization (Diboll and Larson, 1966; Márton et al., 2005). Subsequently one pollen tube first targets the filiform apparatus, a thickened area of secondary cell wall material generated via plasma membrane invaginations of both synergid cells, and then releases its contents explosively into the receptive synergid cell (Higashiyama et al., 2000; Punwani and Drews, 2008; Dresselhaus and Márton, 2009).
Already 50 years ago it was postulated that in lily chemotropic factors secreted from the micropyle are responsible for the final stage of pollen tube guidance (Rosen, 1961; Welk et al., 1965). Initially, the nucellus or adjacent integument cells were discussed as the source of such factors. Later the embryo sac itself was most widely considered as the source of the attractants, and in 1964 van der Pluijm suggested that the synergid cells probably secrete the attractants (van der Pluijm, 1964). This hypothesis was later supported by genetic, cell ablation and in vitro investigations, using excised A. thaliana, T. fournieri and maize ovules, respectively, showing that micropylar or short-range pollen tube guidance appears to be governed by species-specific chemotaxis signals, secreted by the synergid cells of the female gametophyte (Higashiyama et al., 2001, 2006; Palanivelu and Preuss, 2006). The maximum distance of attraction in vitro is in the range of about 200 µm in T. fournieri (Higashiyama and Hamamura, 2008) and 100–150 µm in Arabidopsis (Kasahara et al., 2005; Palanivelu and Preuss, 2006; Stewman et al., 2010), whereas a distance of about 100 µm has been observed in maize (Márton et al., 2005; Márton and Dresselhaus, 2010). Laser-assisted cell ablation experiments using Torenia ovules showed that in this species the synergid cells are necessary and sufficient to attract pollen tubes and that neither the egg cell nor the central cell seem to be involved in this process or are capable of generating sufficient amounts of guidance molecule(s). However, in maize, the egg cell seems to contribute to the production of the attraction signal in addition to the synergid cells (Márton et al., 2005). In Arabidopsis, even the central cell has been suggested to play a critical role in pollen tube guidance (Chen et al., 2007).
Hitherto, only ZmEA1 in maize (Márton et al., 2005; Dresselhaus and Márton, 2009) and LURE1 as well as LURE2 in Torenia (Okuda et al., 2009) have been identified as signalling molecules fulfilling the requisites of female gametophyte-derived pollen tube attractants. ZmEA1 (Zea mays Egg Apparatus 1) is specifically expressed in egg and synergid cells and encodes a polymorphic precursor protein (94 amino acids), which was shown to be secreted from the egg apparatus towards the cell walls of micropylar nucellar cells. In ZmEA1 knock-down plants, pollen tubes arrived at the micropylar cone up to a distance of about 100 µm, but did not penetrate the intercellular space of micropylar nucellus cells, suggesting a role for ZmEA1 in short-range gametophytic pollen tube guidance (Márton et al., 2005). This phenotype is identical to the one described above for maize ovules lacking female gametophytes (Lausser et al., 2010), indicating that ZmEA1 might represent the sole pollen tube attractant in maize. A 49 amino acid oligopeptide, corresponding to the predicted mature N-terminally cleaved ZmEA1 protein, has been shown to attract maize pollen tubes directly at a low concentration (<10 µm) in vitro (Dresselhaus and Márton, 2009; Márton and Dresselhaus, 2010). BLASTP searches revealed the presence of several ZmEA1-related genes encoding small proteins in both monocots and eudicot species that share a highly conserved C-terminal domain, the EA1-box (Márton et al., 2005; Fig. 3). All grasses analysed contain a few conserved EA1-like proteins (EALs) consisting of a P-, EA1- and A-box (Fig. 3). Based on the protein sequence it is not possible to predict the ZmEA1 orthologue in other grass species. Grass EAL proteins display, however, a completely different overall protein structure from Arabidopsis EAL proteins (M. L. Márton, unpubl. res.), which have been misclassified (Gray-Mitsumune and Matton, 2006). We suggest using the term EAC (EA1-box containing) for proteins in both dicots and monocots that contain the EA1-box, but otherwise contain a different protein architecture. According to in silico analysis, all EAL proteins are predicted to be either secreted or contain transmembrane motifs. Predictions localize them to the cell surface or to inner cell membranes. It will now be a task for future investigations to determine if all EAL proteins are involved in pollen tube guidance or in other signalling processes during plant development, and whether EAL proteins can be used to increase interspecific fertilization rates in grasses.
Using novel cell biological tools including male germline-specific marker lines (fluorescent proteins expressed in generative pollen cell and both sperm cells), the behaviour of sperm cells during pollen tube growth, delivery and double fertilization can now be visualized in vivo and studied at the molecular and biochemical level. Without the marker lines described below, a precise description of the fertilization process has not been possible and could only recently be carried out (Berger et al., 2008; Aw et al., 2010). A large number of genes expressed in the male germline are now known in Arabidopsis, tobacco and lily, and have been used to visualize and to study sperm cells in these model plants and to generate, for example, lines producing single sperm-like cells to investigate questions which could hitherto not be experimentally approached, e.g. the occurrence of preferential fertilization (Frank and Johnson, 2009; Ingouff et al., 2009; Borg and Twell, 2010; Oh et al., 2010a). Another example of such new tools is a microspore-specific promoter, which was used to express a green fluorescent protein (GFP)–α-tubulin fusion protein (GFP–TUA6) to visualize all major male germline microtubule arrays in tobacco. These lines can also be used for mass isolation of generative and sperm cells using fluorescence-activated cell sorting (FACS) (Borges et al., 2008; Oh et al., 2010b).
Similar tools have been missing for grasses. Only recently, a male germline-specific marker was reported in maize, allowing study of the sperm behaviour during pollen germination and tube growth, as well as mass isolation using FACS for molecular and biochemical studies (Kliwer and Dresselhaus, 2010). Similar to the approach taken in Arabidopsis, an α-tubulin gene was identified, whose expression during microgametogenesis (pollen development) and pollen tube growth is restricted to the male germline. This line allows visualization of germline cells and was used to visualize and measure the microtubule-rich cytoplasmic tail-like extensions at both sperm cell poles and the position of their nuclei during pollen tube growth in vivo (Fig. 4). Moreover, studies on sperm delivery and double fertilization in vivo become possible and questions such as the occurrence of preferential fertilization can also be addressed in a grass species, if, for example, the promoter of male germline-specific α-tubulin genes driving a photo-switchable GFP variant is used to label either the leading or the trailing sperm cell.
In eudicots, sperm cells have been reported to travel as a male germ unit (MGU) (Dumas et al., 1985). Similarly, a leading sperm cell seems initially also to be associated with the vegetative tube cell during pollen germination in maize (Kliwer and Dresselhaus, 2010). Sperm cell movement in maize and probably other fast growing grass pollen tubes is highly dynamic. Sperm cells may change position relative to each other twice within <15 min (Fig. 4), and movement rates (we have measured on average 650 µm s−1) essentially exceed pollen tube growth rates by more than five times in vitro (Kliwer and Dresselhaus, 2010). During pollen tube growth maize sperm nuclei appear separated by up to 150 µm from each other and it was thus concluded that grass sperm cells are not connected (Heslop-Harrison et al., 1985). In marked contrast, in Arabidopsis, for example, usually the vegetative nucleus takes the lead followed by the leading and the trailing sperm cells without changing their positions (Ge et al., 2011). Recent data now indicate that sperm cells are also always connected during pollen tube growth in maize and α-tubulin seems to be involved in the connection or its stabilization (Kliwer and Dresselhaus, 2010). The analysis of additional cellular marker lines, which are now available via the maize CELL Genomics DB (Mohanty et al., 2009), some of which may be expressed in the germline cells and vegetative tube cell, will provide additional tools for cellular and functional studies of sperm dynamics and behaviour.
During the final stage of pollen tube growth, extensive communication is required between the male (pollen tube) and female gametophyte for successful sperm delivery (Dresselhaus, 2006; Punwani and Drews, 2008; Dresselhaus and Márton, 2009). When sperm cells reach the embryo sac, pollen tube growth is arrested; it bursts and releases both sperm cells into the receptive synergid cell. This process is collectively referred to as pollen tube perception (Weterings and Russell, 2004). Until now, almost all our knowledge about the molecular players involved in sperm delivery has been derived from investigations using Arabidopsis as a model. Various RLKs and glycosylphosphatidylinositol (GPI) anchor-containing proteins such as LORELEI (LRE) (Capron et al., 2008; Tsukamoto et al., 2010), which are localized to the synergid cell surface regulating sperm cell discharge, have been discovered. Signalling of the RLK FERONIA/SIRENE (FER/SIR) is required for female gametophyte-regulated pollen tube growth arrest and reception via the synergid cell (Escobar-Restrepo et al., 2007), while pollen-expressed FER homologous protein ANXUR1 and ANXUR2 (ANX1 and ANX2) activities are discussed as controlling the timing and rupture of pollen tube discharge on the male side (Boisson-Dernier et al., 2009; Miyazaki et al., 2009). In Arabidopsis, it was also shown that degeneration of the receptive synergid cell initiates after the pollen tube arrives at the female gametophyte but prior to pollen tube discharge (Sandaklie-Nikolova et al., 2007; Punwani and Drews, 2008). Mutants of FER, LRE and AMC (Abstinence by Mutual Consent; Boisson-Dernier et al., 2008) fail to induce synergid cell degeneration and repulsion of supernumary pollen tubes, instead showing overgrowth of pollen tubes inside the female gametophyte. Homologues of RLKs such as FER are also expressed in the egg apparatus of maize (N. G. Krohn and T. Dresselhaus, unpubl. res.), but functional studies are lacking to date.
A large step towards the understanding of pollen tube burst has recently been achieved by the discovery that the egg apparatus secreted polymorphic defensin-like protein (DEFL) ZmES4 induces pollen tube rupture in maize via modulating the activity of the intrinsic pollen tube-expressed rectifying potassium channel KZM1 in a species-specific manner (Amien et al., 2010). Following ZmES4 application in vitro, it was shown that KZM1 opens at physiological membrane potentials and closes after wash-out. In summary, it was suggested that vesicles containing ZmES4 are released from the synergid cells upon male–female gametophyte signalling to act on pollen tube tips which have arrived. Subsequent interaction between ZmES4 and KZM1 results in channel opening and K+ influx, followed by water uptake, which likely culminates in osmotic pollen tube burst. Whether the observed depolarization of the pollen tube membrane potential results in the activation of unknown signalling cascades and whether ZmES4 also acts on other membrane transporters, e.g. Ca2+ channels, remains to be shown. Until recently, Ca2+ and not K+ has been considered as the major ion regulating pollen tube growth and perception. Ca2+ was already shown 50 years ago to play a key role in pollen tube germination and growth (Brewbaker and Kwak, 1963; Dumas and Gaude, 2006; Ge et al., 2009). In vitro and in vivo imaging techniques in tobacco, lily and Arabidopsis further demonstrated that growing pollen tubes exhibit a tip-focused Ca2+ gradient and oscillations of cytosolic Ca2+ occur, although oscillations seem less regular in vivo (e.g. Iwano et al., 2009; Michard et al., 2009). In Arabidopsis it has been shown that the pollen tube plasma membrane Ca2+ pump ACA9 is required for pollen tube growth and sperm discharge (Schiott et al., 2004). It remains to be shown whether Ca2+, K+ or both play a role in sperm release in grasses.
In summary, the first tools are now emerging to investigate pollen tube perception in grasses using maize as a model. Small polymorphic cysteine-rich proteins (CRPs) of the DEFL sub-class play a key role in pollen tube burst (Amien et al., 2010). CRPs of the DEFL sub-class are also secreted from the egg apparatus of Arabidopsis (Punwani et al., 2008), indicating that the molecular mechanisms regulating pollen tube perception might be conserved in angiosperms.
With a focus on maize, we discussed various pre-zygotic hybridization barriers in grasses. These are considered to be the major hybridization controls in nature (Arnold and Hodges, 1995; Widmer et al., 2009) and thus contribute to species isolation and speciation (Rieseberg and Willis, 2007). Maize is especially suited to study these processes in more detail as CI is among the most effective isolation barriers to restrict gene flow between populations, and in particular UCI represents the first step towards speciation. UCI occurs naturally between maize (Z. mays ssp. mays) and its subspecies teosinte (Z. mays ssp. mexicana or ssp. parviglumis) as well as among maize inbred lines, facilitating a search for speciation genes in this species. It will be exciting to find out whether CI and UCI mechanisms are similar to SI in eudicot model plant families such as the Brassicaceae, Papaveraceae or Solanaceae, where key molecular players have been discovered and major principles of the various pollination mechanisms have been disclosed during the past 15 years.
We therefore propose to use maize as a model for the economically most important plant family, the Poaceae, not only because of its genetically well described UCI loci, but also because maize has advantages over other grass model plants such as rice and Brachypodium. Maize generates the biggest pollen grains that are easily accessible in huge quantities, making it an excellent model to study pollen development as well as tube growth inside the sporophytic stigma and ovary tissues at the cellular, molecular and biochemical level. A reference genome of the inbred line B73 is finally available (Schnable et al., 2009) and can be considered as a milestone in maize research. Many additional genomes of further inbred lines will be available soon (Belo et al., 2010), enabling study of the evolution of reproductive gene polymorphisms in this species. Moreover, compared with most grass species, Agrobacterium-mediated gene transfer methods in maize have been essentially improved over the years (Ishida et al., 2007) and are routine in many plant biotech companies. Additionally, maize transformation is offered as a service for academia (e.g. www.agron.iastate.edu/ptf or plantsci.missouri.edu/muptcf). Many cellular marker lines are now available (maize.jcvi.org/cellgenomics) to study cellular processes including pollen development, pollination and fertilization mechanisms. A major problem with maize and other grasses is the lack of comparable gene insertion databases, available for eudicot model species such as Arabidopsis. Ongoing current efforts in the maize community aim to establish a transposon mutagenesis library for the purpose of targeted selection of gene knockouts (mtm.cshl.edu), which will provide a very valuable additional tool for functional analysis of candidate reproduction genes in this important plant species. Novel deep-sequencing methods will significantly accelerate the identification of transposon insertions (Williams-Carrier et al., 2010). Similar tools are not available for almost all other grass species, making maize an ideal Poaceae model to study CI and species isolation involving studies on pollen tube growth and guidance as well as sperm delivery.
We thank Ulrich Hammes for critical comments on the manuscript and Irina Kliwer for monitoring sperm cell dynamics. This work was supported by grants of the University Bavaria e.V. to A.L. and the German Research Foundation (DFG grant number MA 3976/1-2) to M.L.M.