Seed samples of 32 species (obligate and facultative sexuals and apomicts of monocots and dicots) were investigated by flow cytometry to reveal the pathway of reproduction. Ten different pathways of seed formation could be reconstructed considering whether the female and/or male gametes were reduced or unreduced, the embryos arose via the zygotic or parthenogenetic route and the endosperm via the pseudogamous or autonomous route. The screen is suited to select sporophytic or gametophytic mutants in sexual species, to identify pure sexual or obligate apomictic genotypes from facultative apomictic species, and to analyze the inheritance of the individual reproductive processes. Corresponding unique results are presented for Arabidopsis, Arabis, Hypericum and Poa. The screen of mature seeds by flow cytometry yielded more information about the reproductive behavior of individual plants than any other available test, and is very useful both in basic research and plant breeding.
The alternation between the sporophytic and gametophytic generation is an essential part of the plant life cycle. Fundamental knowledge of the molecular regulation of the reproduction processes in plants has expanded rapidly during the last decade (for comprehensive reviews see Drews et al. 1998; Goldberg et al. 1989; Grossniklaus & Schneitz 1998; Meinke 1995; Mordhorst et al. 1997). Although most angiosperm species reproduce sexually, asexual seed formation is also widespread and has recently become an important topic both for science and the seed industry (Jefferson & Bicknell 1996; Koltunow 1993; Vielle Calzada et al. 1996). Extensive international research programmes for harnessing apomixis in crop plants are already in progress (reviewed in Matzk et al. 1997). Two main strategies are pursued: the mutative induction of apomixis in sexual model systems, with subsequent transfer of the trait into crops, and the introduction of apomixis from natural apomictic species into sexual crops.
Efficient screening methods for the processes of apomictic versus sexual seed formation are essential for studies of evolution, inheritance and engineering of apomixis as well as for discrimination of mutants associated with apomixis in sexual plants (Czapik 1994; Falque et al. 1998; Mazzucato et al. 1996; Pessino et al. 1999). Such screens are experimentally difficult in angiosperms because the gametophytic generation comprises only a few cells and the ovule tissue, in which the events of megasporogenesis, megagametogenesis and embryo formation occur, is deeply embedded within carpels and therefore hardly accessible (Reiser & Fischer 1993).
Very time consuming serial microdissections of ovaries or progeny tests for maternal/aberrant plants were traditionally employed to assess the mode of reproduction in facultative apomictic species, but these methods are too laborious for routine analyses of large numbers of individual plants. The so-called auxin test (Matzk 1991a; Matzk 1991b) allowed proof of autonomous embryo formation and discrimination of individual plants regarding their capability for parthenogenetic versus fertilization-dependent embryo formation (pseudogamy). This rapid and accurate method is working very well in Pooideae species. In Panicoideae the ‘clearing technique’ is suitable to detect apospory (Herr 1971; Young et al. 1979). Aposporous embryo sacs of several Panicoideae species do not have antipodals, whereas the antipodal apparatus is easily recognizable in meiotic embryo sacs after clearing the ovules (Burson 1997; Chen & Kozono 1994; Hanna et al. 1973; Sherwood et al. 1994).
Depending on whether the embryo sacs are reduced or unreduced and whether or not the egg and/or central cells are fertilized (by reduced or unreduced male gametes), different ploidy levels occur in the nuclei of seed cells as shown in Fig. 1. Diploid sexual plants form a diploid embryo and triploid endosperm. Triploid embryos result from unreduced male or female gametes together with tetraploid or pentaploid endosperms, respectively. But other deviations from the normal sexual pathway also alter the ploidy of embryo and/or endosperm cells. Therefore, reproductive events can be reconstructed from the interrelationship between the DNA contents of nuclei of embryo and endosperm cells.
We have now tested whether or not mature and dry seeds of monocots and dicots are suitable for flow cytometric analyses without any specific pretreatment and without separation of the target tissue. A prerequisite for this approach was that enough intact endosperm cells were still present in mature seeds. The aim was to develop a simple and efficient screening method that identifies simultaneously different reproductive pathways in angiosperms based on the proportional DNA content of embryo and endosperm nuclei, notwithstanding the real ploidy level.
Sexual monocots and specific cases
As expected, seed samples of the sexual species Poa annua, P. supina, P. trivialis, Hordeum vulgare, Triticum aestivum, Oryza sativa, Pennisetum americanum and Zea mays (compare Tables 1 and 2) yielded a high 2C embryo peak and a smaller 3C endosperm peak (Figs 2 and 3). The 2C values of the embryo and 3C values of the endosperm result from the fertilization of the egg cell and the central cell (containing two polar nuclei) of reduced embryo sacs with reduced male gametes, respectively. The seed coat did not contain intact nuclei, as proved in wheat.
Table 1. . Compilation of data from the flow cytometric seed screen in sexuals and in aposporous, diplosporous, pseudogamous or autonomous apomicts of monocots and dicots
Often endopolyploidization occured in endosperm as well as in embryo cells and led to additional peaks with multiple duplications of the basic peak values (e.g. Figures 2i and 3a,c). As these additional peaks are not relevant with respect to the mode of reproduction, they were ignored.
The tri-mutant of barley that produced about 50% unreduced embryo sacs (Finch & Bennett 1979) was characterized by an additional 5C endosperm peak (Fig. 3b). The parthenogenetic wheat line ‘caudata-Salmon’ characterized by 90% autonomous and 20% twin embryo formation (Matzk et al. 1995) yielded, in addition to the common 2C embryo and 3C endosperm peak, a high 1C peak (embryo), indicating haploid parthenogenesis (Fig. 3d–f).
If the embryos are needed to generate progenies, the C value of the endosperm may be determined even from the half of the seed without the embryo, while the C value of the embryo may be inferred from that of the seedling after germination of the second half. This was demonstrated for wheat and Poa pratensis.
Even from normal wheat flour the C value of endosperm nuclei, and from wheat bran the C values of embryo and endosperm cells, could be determined by flow cytometry. This may help to detect an illegitimate use of apomictic wheat in the future.
In diplosporous and aposporous Poa species, a low 5C peak (endosperm) was always found together with a high 2C peak (embryo). The 5C endosperm arose after fusion of the two polar nuclei of unreduced embryo sacs with one reduced male gamete (pseudogamy). The low 5C peak is evidence of unreduced embryo sacs. If it is combined with a high 2C peak, the latter results from the autonomous development of the unreduced egg cell (apomictic seed formation). Poa alpina, P. ampla, P. nemoralis (Fig. 2c) and P. pratensis (Fig. 2e) showed, in addition to the low 5C, a 3C endosperm peak originating from reduced embryo sacs. 2C embryo and 3C + 5C endosperm peaks are typical of pseudogamous facultative apomicts. Only P. palustris, an obligate pseudogamous apomict, exclusively yielded 2C and 5C peaks (Fig. 2b).
Poa pratensis as a facultative apomict showed high variability of peaks (Fig. 2d,e). Analyses of numerous cultivars and F1 plants from crosses between sexual and apomictic genotypes revealed five routes of seed formation, including parthenogenetic development of reduced egg cells and fertilization of unreduced egg cells. The different processes of apomixis could be separated. The results of flow cytometric analyses corresponded with results of the auxin test as well as cytological studies and morphological progeny tests.
For Panicum maximum (Fig. 2f) and Pennisetum ciliare, this screen could not be used to differentiate between sexual and apomictic genotypes (Tables 1 and 2). This was due to the fact that several aposporous Panicoideae species form an embryo sac with four nuclei and only one unreduced polar nucleus. Therefore, the endosperm cells of both the aposporous and sexual plants yielded 3C values. However, genotypes expressing only parthenogenesis or apomeiosis should be detectable by the presence of a 1C peak and a high 3C embryo peak, respectively.
The results for Paspalum dilatatum and Pasp. cromyorrhizon were less clear. The endosperm peaks could not be determined precisely (Table 1). For Pasp. simplex, sexual plants yielded 2C and 3C peaks as expected (Fig. 2g), and the apomictic plants 2C, 3C and 5C peaks (Fig. 2h). The latter plants, previously considered to be obligate pseudogamous apomicts on the basis of other tests (Pupilli et al. 1997), should have only a 5C endosperm resulting from unreduced embryo sacs with eight nuclei (two unreduced polar nuclei fertilized with a reduced sperm nucleus). According to our data, the tested plants were either facultative apomicts/sexuals or the obligate apomicts form embryo sacs with eight as well as four nuclei (with a single unreduced polar nucleus).
The described procedure, using bulked seeds of an individual plant, is suitable to characterize the plant as obligate or facultative for different pathways of reproduction. However, a quantification of the processes in facultative plants remains vague. For quantitative studies, about 50–100 samples of single seeds can be analyzed in Poa pratensis. More efficient is the use of 25–50 samples of two seeds per plant. Each pair of seeds with only one embryo and endosperm peak arose via the same reproductive route; if two embryo and/or endosperm peaks occur in the histograms, the seeds arose by different routes. Hence, the actual degree of sexual or apomictic processes may be estimated rapidly.
Beta vulgaris, Brassica napus, Medicago sativa and Arabidopsis thaliana are all of interest for introducing apomixis. The wild-type plants showed the expected histograms with 2C embryo and 3C endosperm peaks, typical of sexual seed formation (Fig. 4a,c,d).
The histogram of triploid Beta vulgaris seeds (Fig. 4b) showed a 3C embryo and a 4C endosperm peak. This indicates that the seeds arose after fertilization of a diploid mother with a tetraploid father plant. The reciprocal cross would have yielded a 5C endosperm. Similar interrelations between embryo and endosperm DNA values are generally expected for sporophytic mutants forming unreduced male or female gametes.
In Arabidopsis thaliana, diploid (Fig. 4d) and tetraploid lines showed high 2C and 4C embryo peaks besides small 3C and 6C endosperm peaks, respectively. With mixtures of defined numbers of diploid and tetraploid seeds (10 : 40 or 40 : 10) the histograms represent two embryo peaks of sizes (counts) corresponding to the respective seed numbers (Fig. 4e,f). This means that the screen would also allow a quantitative estimate of facultative reproductive processes.
In Hieracium pilosella and Taraxacum officinale (Fig. 5a), autonomous apomixis was demonstrated by the presence of clear 2C embryo and 4C endosperm peaks. Potentilla argentea and Ranunculus auricomus did not yield clear peak differentiation (Table 1).
Arabis holboellii showed high variability regarding the relation of DNA contents of embryo and endosperm (path of seed formation) and the peak position (ploidy). Most frequently, unreduced embryo sacs were formed. Embryo and endosperm formation occured autonomously and/or after fertilization with reduced and unreduced male gametes (Fig. 5b,c).
A large number of geographically widely distributed ecotypes was analyzed in Hypericum perforatum. The results showed a situation similar to that observed in Poa pratensis. The tested seed samples originated mainly from facultative apomictic/sexual (32 samples; Fig. 5d) and facultative meiotic–parthenogenetic/apomictic/sexual genotypes (28 samples; Fig. 5g). Additionally, obligate apomictic (four samples; Fig. 5e), obligate sexual (three samples; Fig. 5f) and aposporous–zygotic/apomictic (one sample; Fig. 5h) plants could be identified. The high quality of histograms (low coefficients of variation for peaks) allowed detection of even small differences of DNA contents. In a few cases, a reduction of chromosome number had apparently occured already in somatic tissue of this tetraploid species (chimeric plants); this was evident from histograms with 1C and 1.5C and/or 2.5C peaks in addition to the expected 2C, 3C and/or 5C peaks (Fig. 5i, without 2.5C). Endopolyploidization was not observed in Hypericum perforatum seeds.
General advantages of the flow cytometric screen of seeds
A novel screen for the route of reproduction has been developed, based on flow cytometric analyses of ploidy levels of embryo and endosperm nuclei from mature seeds. This screen worked accurately and rapidly using standard procedures of flow cytometry in 25 out of 32 tested monocot and dicot species. In seven cases preparation and/or staining needs to be improved for clear differentiation of the embryo and endosperm DNA contents.
The method described can be used independently of the size and construction of the seeds. Even seeds that consist mainly of starch (Poaceae), maternal tissue (Beta vulgaris) or cotyledons (Arabidopsis, Brassica) contained enough intact endosperm cells (aleuron) to determine the DNA content of their nuclei. Seed samples of any plant of interest may be used; the development of specific model systems or genetic lines is not required.
In any case, the amount of nuclei is much lower from endosperm than from the embryo. It is therefore possible to differentiate between embryo and endosperm DNA peaks by their height and position within the histograms. The 3C embryo peaks of the so-called BIII-hybrid embryos, derived from fertilized unreduced egg cells, can also be distinguished from 3C endosperm peaks by their much higher amount of nuclei. The C values of embryo and endosperm nuclei can even be detected from single seeds of Poa pratensis, Triticum aestivum and Zea mays. Moreover, twin embryos (reduced and unreduced) have been identified in single seeds of wheat.
The formation of reduced or unreduced megaspores, zygotic or parthenogenetic embryos, autonomous or pseudogamous endosperms and reduced or unreduced male gametes can be discriminated simultaneously by the flow cytometric seed screen. Ten routes of reproduction were identified in the experiments (Table 2). This method is cheaper and faster, supplies more information, and is affected with a lower error probability than other screens that have been used previously and require separate analyses of the different processes. Moreover, mixoploidy of the somatic cells, which differentiate into the micro- and macrospore mother cells, may result in additional peaks with a specific relation of their C values (peak index) in seed samples, as demonstrated for Hypericum perforatum. The screen will facilitate the breeding process in apomictic crops, introduction of apomixis into sexual crops and analyses of mutations affecting specific steps of female gametophyte and embryo formation in sexual plants. Only in a few cases, the new method is not suited to discriminate between different reproductive processes, e.g. the normal meiotic and an aposporic embryo sac formation of the Polygonum and Panicum type, respectively (both with 2C embryo and 3C endosperm), or if high endopolyploidization of embryo cells and low rates of unreduced male gametes occur (both yield 4C nuclei). It can be expected that the experimental difficulties arising from inadequate DNA staining in certain species can be overcome in future.
Conclusions regarding individual species
Arabidopsis thaliana is generally highly suitable for flow cytometry, as shown previously by differentiation between different trisomic plants (Samoylova et al. 1996). By use of the seed screen it is possible to identify genotypes that produce unreduced male gametes (Matzk et al. unpublished results). The possibility to screen for embryos with different DNA contents within a seed sample and to estimate their ratios has been demonstrated.
Arabidopsis thaliana is a model system for the induction of apomixis within a sexual species. The strategy is based on the selection of seed set without pollination in specific male sterile lines after mutagen treatment (Chaudhury et al. 1997; Ramulu et al. 1998). At least two mutations (for apomeiosis and parthenogenesis) must be induced simultaneously to make this approach successful. By use of our screen the desired mutations may be selected separately in normal fertile lines. The screen might even detect mutants characterized by autonomous endosperm formation, such as FIE, FIS and MEDEA (Chaudhury et al. 1997; Grossniklaus et al. 1998;Ohad et al. 1999) and other non-lethal mutations that affect specific steps of female and male gametophyte development and subsequently alter the ploidy relations in seed cells compared with the wild-type.
Additionally, the procedure is useful to analyze the mode of reproduction in natural apomictic species. The obligate autonomous seed formation in Taraxacum officinale (Falque et al. 1998) has been confirmed in our experiments by the exclusive occurence of 2C embryo and 4C endosperm peaks. The facultative apomixis in Hieracium pilosella (Koltunow et al. 1998) was also confirmed and additionally several obligately apomictic accessions were found.
Arabis holboellii, described as a facultative apomictic species also comprising diploid apomictic strains that are very rare in nature (Böcher 1951), is closely related to Arabidopsis and Brassica napus. Therefore, efforts are being made to use it as a natural apomictic model species. Knowledge about the variability/stability of the single processes of apomixis in this species is limited (Böcher 1951; Roy 1995). The flow cytometric seed screen revealed a high plasticity of the reproductive system of this species. Autonomous and pseudogamous endosperm formation and reduced and unreduced female and male gametes often occured together.
Hypericum perforatum is a medicinal plant used for production of antiviral and sedative drugs. For efficient breeding, recently initiated in several institutions and seed companies, more data about the mode of reproduction are needed. Previous data (Noack 1939; Noack 1941) could be confirmed and extended with the new screen. Mixoploidy occured in seed-propagated plants. This was previously observed only after in vitro regeneration (Brutovskáet al. 1998). For the first time, completely sexual and completely apomictic genotypes were selected. A breeding scheme, comparable to that developed for Poa pratensis (Matzk 1991b), in combination with the new screening method is recommended for Hypericum perforatum breeding.
Poa pratensis has been described as a facultatively aposporous and pseudogamous apomict (Müntzing 1940). It is used as an important forage and turf grass as well as a model species for genetic analysis and manipulation of apomixis (Barcaccia et al. 1997a; Matzk et al. 1997; Matzk 1991b). With the new screen, five distinct routes of seed formation were reconstructed in this species and for the first time genotypes expressing only single processes of apomixis could be selected. Contrary to earlier assumptions (Barcaccia et al. 1997a; Matzk et al. 1997), apospory and parthenogenesis were not closely linked.
The C values observed in seeds from sexual genotypes of Paspalum simplex corresponded with the expectation, whereas the C values from apomictic genotypes disagreed with previous results. Further studies should be focused on questions regarding whether obligate apomicts exclusively form embryo sacs of the Hieracium-type (eight nuclei) or also those of the Panicum-type (four nuclei), and whether or not these genotypes are indeed obligate apomicts.
Conclusions regarding apomixis projects
The advantages of the flow cytometric seed screen and the first results presented here should stimulate re-evaluation of some former results and projects in progress. The observed high variability regarding the mode of seed formation and the frequent separate occurence of the different processes of apomixis, which were shown for several species such as Arabis holboellii, Hypericum perforatum and Poa pratensis, disagree with a simple genetic control of apomictic seed formation as proposed previously (Grimanelli et al. 1998; Ozias-Akins et al. 1998; Savidan 1982).
Some previous interpretations of the evolution and inheritance of apomixis may be incorrect because only one component and not the entire process was considered. Conclusions should be drawn only for those processes that are screened: the auxin test only allows conclusions regarding parthenogenesis, and the clearing technique in Panicoideae only with respect to apomeiosis and not apomixis in general. The flow cytometric seed screen avoids such misinterpretations.
Because apomixis excludes recombination, a close genetic linkage (as a gene cluster) of the single elements provides no selective advantage in evolution. The lack of linkage of apomixis genes may even be an advantage because it offers a higher potential for variation and adaptability within the populations, for instance via ‘haploid’ parthenogenesis and BIII-hybrids (unreduced egg cells fertilized). For the approach of introduction of apomixis into sexual species, the separate isolation and transformation of the corresponding genes from natural apomicts appears to be even easier than engineering a complex of genes. If control by two or more independent genes is the rule, separate markers for each component would be needed, but this is not considered in most marker-based selection techniques for apomixis that are in progress for various species. In these cases, the use of the seed screen would be advantageous for a correct classification of the seed formation in the basic material.
Seed samples that originated from open pollinated individuals or bulked plants of numerous sexual and apomictic monocotyledonous and dicotyledonous species (Table 1) were analyzed by flow cytometry. For each species, three to more than 50 seed samples from botanical gardens, collected ecotypes, cultivars, F1 hybrids or specific experimental lines were used. Depending on seed size, five (e.g. Zea, Beta) to 50 seeds (e.g. Arabidopsis, Hypericum) per sample were chopped with a razor blade in DAPI (4′,6-diamidino-2-phenylindole) staining buffer (DNA staining solution from Partec, Münster, Germany). The extracts were filtered (30 μm) and stored on ice until measurement.
The DNA content of nuclei (C value) was measured using a FacstarPLUS (Becton-Dickinson) flow cytometer, but the Ploidy Analyzer (Partec) was also suitable. The peak position representing the 2C nuclear DNA content of embryo cells was adjusted for each species to channel position 100 or 200 on the linear abscissa scale by changing the voltage of the photomultiplier. Noise signals derived from cellular debris were eliminated with the Facstar machine by gating the target regions in the fluorescence/side scatter dot plot. Additionally, histograms in a logarithmic scale were used for convenient presentation in order to compare the amounts of nuclei of various peaks and to show that no peaks were present outside the linear scale. For each sample, 10 000 nuclei were measured. The so-called peak index was defined as the DNA content of embryo cells in relation to that of endosperm cells. As the different pathways of seed formation were characterized by these specific relations of DNA contents between embryo and endosperm cells and not by the absolute DNA contents, an additional internal standard was not required. The results of this screen were compared with data from other tests, such as cytological studies, the progeny test and/or the auxin test.
The authors thank F. Pupilli, CNR Perugia (Italia), and C.L. Quarin, IBONE Corrientes (Argentina), for the kind supply of defined sexual and apomictic seed samples of Paspalum species, and the editor and two anonymous reviewers for helpful suggestions. A. Meister thanks the ‘Fonds der Chemischen Industrie’ for financial support of the flow cytometric analysis.