Mendel used hawkweeds and other plants to verify the laws of inheritance he discovered using Pisum. Trait segregation was not evident in hawkweeds because many form seeds asexually by apomixis. Meiosis does not occur during female gametophyte formation and the mitotically formed embryo sacs do not require fertilization for seed development. The resulting progeny retain a maternal genotype. Hawkweeds in Hieracium subgenus Pilosella form mitotic embryo sacs by apospory. The initiation of sexual reproduction is required to stimulate apospory in ovules and to promote the function of the dominant locus, LOSS OF APOMEIOSIS, which stimulates the differentiation of somatic aposporous initial (AI) cells near sexually programmed cells. As AI cells undergo nuclear mitosis the sexual pathway terminates. The function of the dominant locus LOSS OF PARTHENOGENESIS in aposporous embryo sacs enables fertilization-independent embryo and endosperm development. Deletion of either locus results in partial reversion to sexual reproduction, and loss of function in both loci results in reversion to sexual development. In these apomicts, sexual reproduction is therefore the default reproductive mode upon which apomixis is superimposed. These loci are unlikely to encode factors critical for sexual reproduction but might recruit the sexual pathway to enable apomixis. Incomplete functional penetrance of these dominant loci is likely to lead to the generation of rare sexual progeny also derived from these facultative apomicts.
Hawkweeds, or Hieracium species, are native to Eurasia, and are taxonomically found in subtribe Hieraciinae in the Asteraceae (Fig. 1). They form multiple florets in a floral head or capitulum (Fig. 1A). Each floret is a perfect flower, containing male and female reproductive organs. The female reproductive organ, the ovary, subtends the corolla and contains a single ovule (Fig. 1A). Hieracium species are highly polymorphic (Fehrer et al., 2007b). In sexual species, female gametophyte development in the ovule begins with meiosis of the megaspore mother cell (mmc) to produce four megaspores. Three of the four megaspores die, and the remaining megaspore undergoes three rounds of nuclear mitosis to form a syncytial embryo sac. Nuclear cellularization and cell differentiation then occur to give rise to an eight-nucleate, seven-celled, Polygonum-type embryo sac, typical of sexual Hieracium species. Double fertilization of the egg and central cell is required for seed initiation (Fig. 1B; Koltunow et al., 1998, 2000). Apomixis in subtribe Hieraciinae occurs by gametophytic apomixis which, in contrast to sexual female gametophyte development, is characterized by the avoidance of meiosis, but retention of mitotic embryo sac development, followed by autonomous or fertilization-independent embryo and endosperm formation (Fig. 1C).
Two different mechanisms of gametophytic apomixis, diplospory and apospory (Fig. 1C) have evolved in two highly diverse Hieracium subgenera, Hieracium and Pilosella, respectively, both of which also contain sexual species (Fig. 1D; Fehrer et al., 2007a). The reproductive and genetic analysis of diplosporous, autonomous apomixis in subgenus Hieracium species has received less attention than aposporous autonomous apomixis in subgenus Pilosella. Reasons for this include low pollen fertility in many species which hampers the development of segregating populations, as apomicts are used as pollen donors in crosses to sexual plants to promote recovery of hybrids. The species which have currently been examined are less amenable to micropropagation and tissue culture, which impedes functional analyses (SDJ and AMGK, unpublished observations).
The high pollen fertility and ease of Agrobacterium-mediated transformation of species in subgenus Pilosella has led to the development of this subgenus as a model system for the molecular analysis of aposporous, autonomous apomixis (Bicknell and Koltunow, 2004). Subgenus Pilosella species have a base chromosome number of n=9. The sexual taxa are diploid and also polyploid, while apomicts are polyploid and also aneuploid, although diploid apomicts can be generated in experimental populations (Bicknell, 1997; Bicknell et al., 2003). Subgenus Pilosella species are self-incompatible (Gadella 1984), but gene flow across ploidy levels is common as the sexual and apomictic species can interbreed, forming allopolyploid hybrids (Fehrer et al., 2007b).
Apomixis is a dominant genetic trait in subgenus Pilosella, and apomicts are facultative, which means that although the majority of progeny are formed asexually and have a maternal genotype, the capacity to form seeds via sexual reproduction is also retained (Koltunow and Bicknell, 2004). A small fraction of progeny in these apomicts arise from meiotically produced eggs. Unreduced female gametes are obviously prevalent in apomictic subgenus Pilosella; however, unreduced male gametes are also observed in sexual and apomictic plants, as are aneuploid pollen grains (Fehrer et al., 2007b).
Here, we reflect briefly upon Mendel's experiments on Hieracium and his findings with respect to our current knowledge of apomixis in hawkweeds. Developmental aspects of apomixis in subgenus Hieracium and subgenus Pilosella are then considered. The remainder of this review summarizes current information concerning the developmental functions of the two dominant loci known to control apomixis in aposporous subgenus Pilosella. It then focuses on the interplay between sexual and apomictic pathways in order to explain how both apomictic and non-maternal progeny types arise from an apomict.
Mendel's experiments on Hieracium
Gregor Mendel's experiments in Pisum led to the establishment of the fundamental laws of genetic inheritance. When his true breeding (homozygous) pea plants were crossed, the F1 hybrids (heterozygotes) were uniform in appearance. Variable progeny were, however, obtained from self-pollinated plants in the F2 generation as a result of trait segregation. He utilized >20 species from various genera to corroborate these finding in pea. He considered Hieracium species in subgenus Pilosella and also Hieracium (then named Euhieracium) to be attractive candidates as parents for initial crosses to generate F1 hybrids because some appeared to already be ‘true breeding’ for their morphological traits. The colour variation of flowers and differences in morphology between species made Hieracium useful material for visually detecting hybrids. Nogler (2006) has provided a detailed interpretive account of the experiments Mendel conducted in particular Hieracium species from his short publication (Mendel, 1869) and an analysis of the 10 letters originally published by (Correns 1905) that Mendel wrote during 1866–1873 to Karl Nägeli, the botanist who specialized in Hieracium.
In summary, when Mendel set up his first crosses with ‘true breeding’ Hieracium plants in 1866, he was unfortunately not privy to the reproductive peculiarities of Hieracium species. He obtained frustratingly different results in Hieracium compared with his Pisum data when he emasculated and crossed, for example, what we now know are two apomictic parents. Hybrids were extremely difficult to obtain. He observed that the F1 progeny mainly retained a maternal morphology and in his letters to Nägeli he stated his concern about his emasculation technique (Correns, 1905). Nevertheless Mendel managed to isolate some hybrids from these crosses. Perplexingly, these F1 hybrids were not morphologically uniform in appearance as a contemporary experimental example in Fig. 2 shows. It is now known that this is because these parents were not at all ‘true breeding’ homozygous plants but they were highly heterozygous apomicts. In complete contrast to his work in pea, the F2 progeny derived from examined F1 plants were predominantly uniform and resembled their parent in morphological appearance (see F2 plants in Fig. 2). Uniform progeny continued to be stably maintained in subsequent generations and he analysed progeny over four generations in a particular experiment (Correns, 1905; Nogler, 2006). These observations in Pisum and Hieracium led, in subsequent years, to the notion of two distinct types of inheritance in plants: the Pisum type where trait segregation occurred, and the Hieracium type where trait segregation was largely absent (Nogler, 2006). The peculiarities of Hieracium were clarified and found to be related to the mode of reproduction these plants possessed when the existence of apomixis was discovered in these plants following emasculation and cytological analyses (Ostenfeld, 1904, 1906; Rosenberg, 1906, 1907; Nogler, 2006).
In retrospect, Mendel's experiments in Hieracium had provided the first evidence for the co-existence of sexual and apomictic reproduction in these apomicts. His data clearly showed that while predominantly maternal progeny were produced in crosses between apomictic subgenus Pilosella species, sexual reproduction was not eliminated as hybrids could be obtained. How apomixis occurs and is controlled in Hieracium and the factors that give rise to sexual and apomictic progeny are considered in the following sections.
Diplosporous apomixis in subgenus Hieracium
Diplosporous apomixis has not been extensively analysed in subgenus Hieracium species. Analyses of 22 apomictic species are currently underway to examine the modes and variation in the events of diplosporous reproduction. Cytological analyses in H. murorum are summarized in Fig. 3A–E. A cell differentiates in the position typically occupied by the mmc in sexually reproducing plants, yet it does not appear to go through meiosis. It undergoes nuclear mitosis to form a syncytial embryo sac (Fig. 3B, C), and the subsequent cellularization of the nuclei results in the formation of an embryo sac with an irregular structure. Figure 3D shows an embryo sac containing three elongated antipodals (top), an early elongated zygote-like embryo, and fused central cell nuclei. Rounded eggs with large vacuoles typical of those found in mature embryo sacs of sexual subgenus Hieracium species examined are difficult to find in H. murorum, implying a rapid transition to embryogenesis (Fig. 3D). Multiple embryos are also observed in ovules, and from their position they appear to originate from synergids (Fig. 3E). Emasculation of H. murorum has confirmed that seed formation is fertilization-independent (AMGK and SDJ, unpublished observations).
In other diplosporous apomictic species the mmc enters meiosis, but the process ceases with the mmc or one of its products undergoing mitotic embryo sac formation. In these cases it seems that there has been a switch from a sexual pathway to an apomictic pathway in the one cell lineage (Koltunow and Grossniklaus, 2003; Tucker and Koltunow, 2009). In subgenus Hieracium species it is currently unknown whether an mmc is specified first and then switches to an apomictic pathway, or if the cell that differentiates in the position of the mmc is already fated to undergo apomixis and is mitotically determined (hence ?mmc? in Fig. 1C). The use of cell type-specific markers is required to resolve this. For example, the presence of callose (a marker correlating with sexual identity) in the walls of the cell initiating diplospory, together with the expression of meiosis regulatory genes, would support mmc identity and a switch to a diplosporous pathway, and this is under investigation.
Aposporous apomixis in subgenus Pilosella
Cytological, genetic, and molecular analyses have primarily been focused on apomixis in plants found in the two divergent chloroplast haplotype groups of Hieracium subgenus Pilosella (Fig. 1D) that undergo apospory. In these apomicts, sexual reproduction begins first with the differentiation of an mmc that undergoes meiosis. The mmc and its tetrad of meiotic products contain callose in their cell walls (Tucker et al., 2001). During these events, one or more aposporous initial (AI) cells differentiate close to sexually determined cells and expand towards them as they undergo nuclear mitosis. The sexual pathway ceases with the degeneration of all meiotic products (Fig. 3F; Koltunow et al., 1998, 2000, 2011; Bicknell and Koltunow, 2004).
Multiple aposporous embryo sacs can form with a frequency that is species dependent (Koltunow et al., 1998, 2000, 2011; Bicknell and Koltunow, 2004). Aposporous embryo sacs develop synergids, one or more egg cells, and a central cell, but antipodals are rarely observed. Once the egg and central cells differentiate, embryo and endosperm formation occurs rapidly thereafter, often prior to flower opening (Fig. 3G–I). This contrasts with the arrest of meiotically derived eggs and central cells resulting from sexual reproduction as they await fertilization to activate seed initiation.
In H. aurantiacum a different mode of aposporous embryo sac formation has been observed where multiple embryo sacs amalgamate into one as cell wall degeneration occurs during embryo sac formation (Koltunow et al., 1998, 2000). There does not appear to be a correlation between the chloroplast haplotype group location of a particular apomictic subgenus Pilosella species and the mode of aposporous embryo sac formation, as species with both modes of aposporous embryo sac formation co-exist in the same haplotype group (Fig. 1D; Koltunow et al., 2011).
Identity of the cells initiating apospory
AI cells in Hieracium, like sexual megaspores that also undergo mitosis to form megagametophytes, do not contain detectable levels of callose in their cell walls (Tucker et al., 2001). AI cells do not express the marker AtSPOROCYTELESS:GUS which in sexual and apomictic subgenus Pilosella species marks the mmc (Tucker et al., 2003). HDMC1, a gene that functions in chromosome pairing during meiosis, is expressed in enlarging mmcs but is not expressed in AI cells (Okada et al., 2007). These data indicate that AI cells do not have mmc identity and are therefore unlikely to be additional mmcs that have differentiated in a nearby location and subsequently switched to an apomictic pathway.
A marker expressed in the functional megaspores of Arabidopsis (Huanca-Mamani et al., 2005) is also absent in the AI cells of apomicts, but it is not expressed in the functional megaspores of sexual Hieracium, indicating that this marker does not have megaspore specificity in Hieracium (Koltunow et al., 2011). There is no doubt that AI cells have the capacity to undergo nuclear mitosis and form an embryo sac soon after differentiation, but the gene expression pathways evident in AI cells as they differentiate remain unknown. A comparison of the gene expression programmes in individual AI cells and early aposporous embryo sacs with somatic ovule cells using RNA isolated from cells harvested by laser capture microdissection is underway (Fig. 4A). Transcriptomic comparisons of expressed genes in these cell types should provide insight into the gene expression programmes evident in differentiating AI cells as they undergo transition into aposporous embryo sacs (Fig. 4B).
Sexual reproduction needs to initiate before apospory can begin
In aposporous subgenus Pilosella species the events of sexual reproduction begin first with the differentiation of the mmc, its entry into meiosis, and formation of a tetrad of megaspores. AI cells differentiate during these events and undergo nuclear mitosis during sexual megaspore degeneration. In order to address whether early sexual events are required for AI cell differentiation, the developmental arrest of meiosis in sexual and apomictic species was targeted using a cytotoxic reporter gene comprising the AtSPOROCYTELESS promoter fused to BARNASE. This promoter drives linked gene expression to mmcs in sexual and apomictic Hieracium but is not expressed in AI cells (Tucker et al., 2003). Meiotic arrest was observed at different stages in ovules of sexual H. pilosella and apomictic H. piloselloides transgenics; however, due to the low level of expression, some ovules were able to progress through meiosis in both sexual and apomictic plants. AI cells did not develop in ovules of sexual transformants at any stage. In apomictic H. piloselloides, AI cells and aposporous embryo sac formation were not observed in ovules where meiotic arrest occurred at the mmc or dyad stage. AI cells formed only when meiosis was able to proceed through to the tetrad stage of megaspore development. In H. piloselloides the initiation of apospory is dependent upon the prior sexual events that enable megaspore tetrad formation in ovules, indicating an early interplay between sexual and apomictic pathways (Koltunow et al., 2011).
Apomixis in subgenus Pilosella is not a completely novel pathway requiring a unique infrastructure
Once the AI cell undergoes its first nuclear division, and the subsequent events of aposporous embryo sac development and autonomous seed formation occur, the spatial and temporal expression of reproductive marker genes in two apomictic species occurs in the same manner as that observed in a sexual subgenus Pilosella species. Thus apomictic and sexual pathways share common genes and regulators during the mitotic events of aposporous embryo sac formation and autonomous seed development (Tucker et al., 2003). Once initiated, the events of aposporous apomixis can be viewed as a modified sexual pathway in a somatic cell lineage whereby embryo sac formation occurs mitotically and the requirement for fertilization to trigger seed initiation is bypassed.
The functions of two dominant, independent loci that control apomixis in H. praealtum
Are the factors controlling apomixis an essential part of the sexual machinery? The events of meiosis and fertilization are avoided in H. praealtum by the actions of two dominant, independent genetic loci, LOSS OF APOMEIOSIS (LOA) and LOSS OF PARTHENOGENESIS (LOP; Catanach et al., 2006; Koltunow et al., 2011). The characterization of a suite of apomixis mutants generated by γ-irradiation in apomictic H. praealtum has enabled a greater understanding of the role these loci play in apomictic development and their relationship with the sexual pathway (Catanach et al., 2006; Koltunow et al., 2011).
LOA functions sporophytically and is required for the differentiation of AI cells and the subsequent suppression of the sexual pathway. LOP acts gametophytically and is required for autonomous embryo and endosperm development, and these functions are tightly linked at the LOP locus (Koltunow et al., 2011). In mutants where LOA is non-functional and the LOP locus is functional (loaLOP), the process of meiosis initiates, AI cells do not form, and the subsequent events of sexual gametophyte development proceed to completion (Fig. 3J). LOP segregates in meiotically reduced embryo sacs. Embryo sacs that do not inherit LOP remain quiescent but can be fertilized to give rise to viable hybrid seeds (Fig. 3K). Embryo sacs that inherit LOP can undergo fertilization-independent seed development and give rise to viable seedling progeny with half the ploidy of the maternal parent (Fig. 3L; Koltunow et al., 2011).
In mutants where LOA is functional and LOP is non-functional (LOAlop), the events of meiosis begin, AI cells differentiate, sexual reproduction is suppressed, and unreduced, aposporous embryo sacs form that are not capable of autonomous seed initiation. They can be fertilized to give rise to viable hybrid seedlings that exhibit a ploidy level higher than that of the maternal parent, indicating that the unreduced eggs are receptive, like meiotically reduced eggs, to fertilization. In mutants where both LOA and LOP are deleted and non-functional (loalop), the plant reverts to sexual reproduction and the apomictic pathway is absent (Fig. 3M; Koltunow et al., 2011).
In aposporous H. praealtum, sexual reproduction is the default pathway and LOA and LOP are not essential for sexual reproduction
Analyses of apomixis mutants has shown that in the absence of LOA and LOP loci sexual reproduction is the default reproductive mode in H. praealtum and therefore these two apomictic loci are not critical for sexual reproduction. They are thus unlikely to encode mutant alleles of essential genes involved in sexual reproduction as some models propose (Koltunow and Grossniklaus, 2003). LOA and LOP are, however, required sequentially to induce apomixis in a somatic cell lineage. They might somehow recruit the sexual machinery that enables mitotic events of embryo sac formation and embryo and endosperm development given that these mitotic apomictic pathway processes share genes and regulators with the sexual pathway (Tucker et al., 2003).
Suppression of the sexual pathway in H. praealtum occurs as a consequence of LOA action as, once AI cells form, the demise of the sexual pathway occurs with the degeneration of the tetrad of megaspores. This might occur indirectly by the enlarging aposporous embryo sacs physically crushing the sexually determined cells as the enlarging aposporous embryo sacs invade the space occupied by sexual cells. Alternatively, it may be a direct function of the LOA locus mediated by the functions of one or more genes that LOA encodes. The outcome of the stimulation of mitotic embryo sac formation by LOA means that the eggs, progenitors of the embryo, and the subsequent generation inherit both LOA and the unlinked LOP locus which functions to trigger fertilization-independent embryo and endosperm formation.
Curiously, in H. praealtum, embryo sacs inheriting LOP are resistant to fertilization. This observation arose from the analysis of mutants and segregating wild-type H. praealtum plants (Koltunow et al., 2011). The reason for this is unclear but may relate to the precocious initiation of egg and central cell proliferation before the flower opens in this species. This phenomenon of LOP-carrying eggs being resistant to fertilization is unlikely to be a feature of all apomictic Hieracium species as other studies have suggested that the unreduced eggs derived via apomictic reproduction can be fertilized to give rise to plants of increased ploidy relative to the maternal parent (Bicknell et al., 2003; Bicknell and Koltunow, 2004; Krahulkova et al., 2009; Fig. 5). Thus in H. praealtum where gametophytes carrying LOP are resistant to fertilization, the chances of preserving the maternal apomictic genotype in progeny are further maximized (Koltunow et al., 2011).
Mature embryo sacs from Hieracium ovules can be manually isolated (Fig. 4C–H) and the method is described in the legend to Fig. 4. The use of transcriptomics to compare gene expression profiles between the embryo sacs of sexual, apomictic, and the described apomixis mutants with partial or full reversion to sexual reproduction should provide important information on pathways stimulated in the presence and absence of fertilization.
How are apomictic and sexual (‘off-type’) progeny generated in apomicts?
Most of the progeny germinating from seeds arising from apomictic subgenus Pilosella species are maternal in genotype and reproduce via apomixis. Seedlings that do not retain the maternal clonal genotype, termed ‘off-types’, also naturally arise from apomictic plants. They have been quantified and characterized using flow cytometry following a series of emasculation and cross-pollination treatments (Bicknell et al., 2003; Bicknell and Koltunow, 2004; Fig. 5). Such ‘off-types’ generally form a very small subset of the progeny, and they can be subdivided into a number of classes. Unreduced eggs in mitotically formed gametophytes of Hieracium can occasionally be fertilized to form seedlings, termed addition hybrids, that have increased ploidy with respect to the maternal parent. Meiotically reduced eggs can form and be fertilized to form hybrids with the same ploidy as the maternal parent. Polyhaploid progeny are also found; these are derived from the autonomous development of a meiotically reduced egg and exhibit half of the ploidy of the maternal parent.
From our current understanding of apomixis in subgenus Pilosella, the loci that control it, and the early interactions between the sexual and apomictic pathways, an explanation for the generation of apomictic and ‘off-type’ progeny can be provided. The scheme summarized in Fig. 5 shows the types of progeny that can be derived from a tetraploid apomict containing a single copy of each dominant LOA and LOP locus both in the absence of fertilization and following fertilization with a tetraploid sexual plant.
In the apomict, the sexual pathway initiates first with mmc formation and entry into meiosis. This enables LOA function and the stimulation of AI cell formation and subsequent suppression of the sexual pathway (Fig. 5A). The mitotic events of aposporous embryo sac formation ensure that both LOA and LOP are inherited in the eggs of the unreduced gametophytes. LOP functions gametophytically to stimulate autonomous embryo and endosperm formation in the embryo sacs. Seedlings arising from this apomictic pathway have the same genotype as the parent and are capable of apomictic reproduction (Fig. 5B). For example, in the tetraploid apomict H. piloselloides, apomictic progeny form 97.05% of the viable seed (Bicknell et al., 2003). The unreduced eggs formed during aposporous gametophyte development in H. piloselloides can be fertilized at low frequency, giving rise to addition hybrids with increased ploidy relative to the maternal parent. These comprise 0.1% of the progeny in H. piloselloides (Bicknell et al., 2003).
As sexual reproduction initiates first in apomicts, and appears to be required for LOA activation, if LOA function is not activated in the ovule and/or LOA fails to suppress meiosis, then sexual female gametophyte formation can proceed unhindered, resulting in segregation of the LOA and LOP loci. Subsequent development of seeds in the presence or absence of fertilization, shown in Fig. 5C, gives rise to additional ‘off-type’ progeny. Depending on the loci the eggs inherit, the progeny may retain the ability to reproduce via apomixis, may have lost the ability to do so and therefore reproduce sexually, or they may have the ability to undergo a partial form of apomixis. In H. piloselloides, the ‘off-types’ arising from meiotically reduced eggs form 2.85% of the viable seed (Bicknell et al., 2003). In other apomictic subgenus Pilosella species, the percentage of apomictic and ‘off-type’ progeny formed varies (Fehrer, 2007b). This may potentially reflect variation in the penetrance of LOA function to successfully stimulate AI formation and suppression of sexual reproduction and similarly variation in the efficiency and timing of LOP function to trigger autonomous seed development.
In retrospect, Mendel must have conducted thousands of crosses in order to obtain the rare hybrids he generated from various cross combinations in his experiments with Hieracium. Functionally, the ‘off-types’ produced via meiosis and fertilization can be viewed as a diverse ‘back-up’ cache of new sexual and apomictic genotypes, providing an opportunity for the apomictic species to evolve if environmental pressures select against the maternal genotype. The ‘off-types’ together with the capacity for sexual and apomictic subgenus Pilosella species to hybridize drive evolution in this subgenus, contributing to the diversity and tremendous variation in form observed in this subgenus (Krahulkova et al., 2009; Koltunow et al., 2011).