Flowering plants have evolved a complex fertilization mechanism involving two male and two female gametes. Double fertilization produces the embryo and the endosperm, which protects the embryo and regulates trophic interactions between the embryo and the mother. Since the discovery of double fertilization at the end of the ninetieth century by Guignard (1899) and Nawaschin (1898), this unique reproductive process has remained rather enigmatic. Direct cytological observations are difficult as the dynamic interactions between relatively small gametes takes place deep in the surrounding maternal tissues. Only in the mid-1960s, electron microscopists revealed that both male and female gametes are devoid of cell wall, allowing gamete interaction and fusion (Jensen 1965; Jensen and Fisher 1968). These cytological descriptions were repeated in various species and accompanied by a significant body of experimental data gathered from mainly two types of experimental approaches. The design of methodologies allowing in vitro fertilization in maize and other species led to important physiological studies, unraveling common calcium mediated signal transduction during fertilization in animals and in plants (reviews by Kranz and Scholten 2008, this issue). Laser dissections of ovules from Torenia established the major role played by synergids in short range attraction of the pollen tube (Higashiyama and Hamamura 2008, this issue). Eventually, species producing large gametes and with specific features facilitated experiments leading to a fairly elaborate description of double fertilization (Weterings and Russell 2004). However hardly any molecular mechanisms directly involved in double-fertilization had been identified and researchers proposed conflicting hypotheses and concepts.

During the past few years, molecular and genetic approaches using Arabidopsis have triggered a considerable renewal of interest in gametophyte and double-fertilization. This special issue commemorating the 20th anniversary of Sexual Plant Reproduction is timely as major findings have been obtained recently, and significant advances now promise rapid new development of this field, essential for breeders and agro biotechnology. Some reviews in this volume summarize the overall contribution from older approaches and other reviews focus on still unsolved problems and new directions that remain to be explored.

The major steps of double fertilization consist successively in (1) the attraction and arrest of the pollen tube, (2) the release of the two male gametes in the degenerated synergid, (3) the migration of male gametes to the two female gametes, (4) gamete recognition and fusion, (5) the fusion of the parental genetic material during karyogamy, (6) re-initiation of the cell cycle, transcription and translation, leading to a new start of the zygotic life. Recently data have been gathered providing some clues on the mechanisms involved in the first two steps. There is currently hardly any knowledge concerning subsequent steps.

One major recent advance in the past few years concerns the origin of pollen tube attraction and the role of synergids (Higashiyama and Hamamura 2008, this issue; Punwani and Drews 2008, this issue). Defects in synergid differentiation in the myb98 mutant cause loss of pollen tube attraction (Kasahara et al. 2005). The sirene and feronia mutants in Arabidopsis have shown that the female gametophyte controls the arrest of the pollen tube required for male gamete release (Huck et al. 2003; Rotman et al. 2003). The FERONIA gene encodes a receptor kinase specifically expressed in synergids (Escobar-Restrepo et al. 2007), suggesting an exchange of signal between the pollen tube and the synergids.

Transcriptomes of male and female gametophytes have been characterized leading to an overall picture of a larger diversity for pollen-specific genes in comparison to embryo sac specific genes. The transcriptome of female gametophytes were obtained from comparisons between wild-types and mutants lacking embryo sacs and more recently from isolated cells isolated through laser microsurgery (Johnston et al. 2007; Jones-Rhoades et al. 2007; Steffen et al. 2007). These gametophytic expression profiles will be essential tools, enabling the characterization of novel genes and gene networks and generation of critical functional mutant lines.

The angiosperm gametes are deprived from a motile apparatus. Whether there is a true migration of male gametes remains unclear since the explosive pollen tube discharge could be sufficient to project each gamete to the female gamete. However, observations of fertilization in vivo support an active migration of male gamete toward each female gamete and further migration of the male nuclei (Higashiyama et al. 1997; Ingouff et al. 2007). Actin coronas around the male gametes have been observed in fixed samples, suggesting that a migratory mechanism based on actomyosin interactions could take place (Márton and Dresselhaus 2008, this issue). Direct confirmation of these early observations need to be obtained. Whether each male gamete is actively directed to each female gamete and how gamete recognition takes place likely play a role in the prevention of polyspermy (Spielman and Scott 2008, this issue). The recent identification of the protein GSC1 expressed specifically on the surface of sperm cells is of particular importance (Mori et al. 2006; von Besser et al. 2006). This protein is conserved across a large spectrum of species and will likely provide an entry point into the molecular mechanisms of gametic interaction. The stage is set for the molecular repertoire of the male gamete to unfold as an important partner to double fertilization (Singh et al. 2008, this issue)

Karyogamy was observed clearly in Arabidopsis both in fixed material (Faure et al. 2002) and in vivo (Ingouff et al. 2007) and is associated with unusual features, suggesting that both parental chromatins do not merge immediately. It was suggested some years ago that the paternal genome is silenced (Vielle-Calzada et al. 2000), although since then several lines of evidence disproved the general impact of this hypothesis (Meyer and Scholten 2007; Weijers et al. 2001). Hence there is probably a range of parental expression of the genome from genes expressed equally from both parental alleles immediately after fertilization, to extreme cases subjected to differential expression according to the parental allele origin. These genes are thus imprinted, reflecting different epigenetic marks acquired or lost during gametogenesis (Gehring et al. 2004; Curtis and Grossniklaus 2008, this issue).

The onset of zygotic gene expression parallels the reinitiation of the cell cycle. Although it is now fairly clearly established that the male gametes are delivered at the G2/M transition (Durbarry et al. 2005), the status of the female gametes remains obscure. Since the initiation of the cell cycle is nearly immediate in the endosperm in comparison to the embryo, it is likely that the female gametes are not arrested at the same stage. What causes the arrest and prevents unwanted onset of development in absence of fertilization has been partially uncovered in the past years. A Polycomb group complex (PgC) controls this arrest in the central cell, whereas the arrest of the egg cell is dominated by another pathway, which may involve the tumour suppressor retinoblastoma (Curtis and Grossniklaus 2008, this issue). However, a cautious study of the cell cycle status of the female gametes based on markers remains to be performed in Arabidopsis.

It is becoming gradually clear that although plant and animal kingdoms diverged more than 1 billion years ago, similar mechanisms govern sexual reproduction in both kingdoms. The review by Márton and Dresselhaus (2008, this issue) outlines some of these parallels. The current idiosyncratic nomenclature used to designate plant reproduction has obscured the parallels that now become apparent between plants and animals. It is likely to be the time to rethink the designation of each actor of the reproductive process such that the literature in the field becomes relevant to a broader readership working in the field of reproductive biology.

Although several plants with particular reproductive features have been essential to perform seminal experiments, the model species Arabidopsis is likely to become the relevant species for genetic dissection of double-fertilization. The recent development of fluorescent markers for both male and female gametes and the availability of transcriptional profiles from isolated tissue will provide powerful tools when combined with functional genomics and advanced genetic screens. Major hurdles still to be face in a more distant future are physiological approaches in vivo, the means to manipulate at will each fertilization event and a clearer knowledge of apomixis related genes and technologies. On the longer term, the main question remains open: why such an unusually complex reproductive system was selected and what part it played in the formidable success of angiosperms on Earth?