Induction of plant somatic embryogenesis in liquid medium

The large scale propagation of plants via somatic embryogenesis, has so far been difficult to achieve. In this thesis research is described leading to embryogenic cell lines that can be maintained for a long period, without loss of genetic stability. It is also described how embryogenic potential of cell lines can be influenced by the addition of specific arabinogalactan-proteins.We consider the large scale production of somatic embryos to consist of five steps; initiation of embryogenic cell lines, proliferation of pro-embryogenic masses (PEMs), formation of embryos, germination of embryos and transfer of germinated embryos to the greenhouse. We have found for three crops, carrot, cyclamen and cucumber, that when the first three steps are performed in liquid medium, embryogenic cell suspensions can be obtained in a very comparable manner. The cell lines produce PEMs, that proliferate at a similar growth rate, for all the three crops, and produce somatic embryos at a high efficiency. The somatic embryos germinate and produce plantlets which grow into mature, fertile plants, again with a high efficiency. Since the first three steps are performed in liquid medium, the process is labour extensive and inexpensive. The major costs in large scale productions will then be associated with the germination of the embryos and the transfer to the greenhouse. The achievement of the research presented in this thesis is that the feasibility of somatic embryogenesis for plant propagation is demonstrated for three different crops, under conditions that preserve genetic stability and embryogenic potential of the cell lines.Essentials for the initiation phase.For the initiation of an embryogenic cell line, directly in liquid medium, an explant has to be chosen, which is able to produce PEMs if incubated in the proper medium. In our experience, an explant derived from a piece of the plant close to the zygotic embryo, both in space and in time, has the best chance of success (Carman 1990, Debeaujon and Branchard 1993, Williams and Maheswaran 1986). Explants from a germinating seed, like root tips from cucumber (Chapter 6) or hypocotyls from cyclamen (Chapter 5), have proven to be very useful, both in our experience and in literature (Carman 1990, Debeaujon and Branchard 1993, Williams and Maheswaran 1986).In initiation, the nutrient composition and the hormone concentrations of the liquid medium in which the explant Was placed, are very important. For each crop we observed a different optimal composition. The ammonium concentration in different media varies greatly, from 2 mM in Gamborgs B5 medium to 25 mM in KK medium. The higher concentrations might be toxic to carrot and cyclamen, but not to cucumber. Carrot and cyclamen could be initiated in B5 medium, but cucumber needed MS medium, which contains 21 mM ammonium, In all cases the media, contained an auxin, 2,4-dichlorophenoxyacetic acid (2,4-D), and usually also kinetin. These hormones were rapidly taken up by the explant (Chapter 5, Fig. 2, Chapter 6, Fig, 1) and we showed that rapid uptake of hormones is likely a prerequisite for induction of embryogenic potential.If root tips or hypocotyls were used, the incubation of the explant resulted in the dispersion of the epidermis and cortex, leaving the vascular bundle and the surrounding cells intact. Probably, the cells surrounding the vascular tissue provide cells able to produce the PEMs. A similar conclusion was drawn by Nadolska-Orczyk and Malepszy (1987), in their study on the initiation of embryogenesis on leaf explants of cucumber. Although these experiments were performed on solid medium, embryo formation was initiated in cells surrounding the vascular tissue.In our experience, PEM formation was achieved within a period of maximal 8 weeks, although differences between crops were observed, carrot always being the fastest and cucumber the slowest. What causes the differences between the three crops in the length of the initiation phase is unknown. PEMs were either formed on the explant or from single cells or cell aggregates in the medium, and were formed in massive amounts at an almost predictable point during the initiation phase.The use of different explants, or media with different nutrients or hormone concentrations could lead to delay, or absence, of PEM formation, resulting in poorly, or non-embryogenic cell lines. In these cases, root formation was often observed. Already in 1966, Halperin reported similar observations in carrot cell lines. In his paper (Halperin 1966) on alternative morphogenetic events in carrot cell suspensions he stated that "Strong circumstantial evidence indicates that the same cells in culture are capable of giving rise to either a rootbearing clump or to an embryo, depending upon the particular chemical environment in which they are grown. The present data suggest that the formation of proembryos depends upon a rapid rate of cell division under conditions which prevent cell enlargement." We have found that this also holds for cyclamen and cucumber cell lines. Indeed, rapid cell division may repress cell enlargement, and thus promote the formation of small cells, which are able to become embryogenic cells.Furthermore we have found that the nutrient composition is equally important as the concentration of the hormones. In a recent paper Preece (1995) concluded "There is an interesting relationship between nutrient salts in the medium and the plant growth regulators (PGR). However, nutrient salts will not wholly substitute for PGR, and neither will PGR completely substitute for nutrient salts. With the improper nutrient medium, chances will be very low that explants will respond as desired, regardless of the PGR and their concentrations tested. Conversely, with the improper PGR type, combination, or concentration, explants will respond poorly if the nutrient salts are less than ideal for that plant genotype or the physiological state of the plant material." Indeed, also in our experience, the type of explant determines which is the ideal combination of nutrients and hormones for obtaining maximal embryogenic potential in the initiation phase. This is further illustrated by the preference of cucumber for MS medium (Chapter 6), while cyclamen (Chapter 5) and carrot prefer B5 medium (de Vries et al. 1988).The induction of embryogenic potential in explants, placed in liquid medium, has been described for only a limited number of plant species. Embryos of cucumber (Ziv and Gadasi 1986), cassava (Raemakers et al. 1993) and melon (Oridate el al. 1992) were obtained, on explants placed in medium containing an auxin. After an induction period in auxin containing medium, the embryos developed on the explants, which were then transferred to hormone free medium. No PEMs were observed, and no embryogenic cell suspensions were formed. This method therefore differed form the one described in this thesis. The importance of the initiation phase was also emphasized for Hevea brasiliensis ( Micheaux-Ferriere and Carron 1989). In this case the length of time of the first two subculture periods on solidified, auxin containing medium, determined whether embryos were formed in the third subculture period, after transfer to auxin-free medium. Although PEMs were not observed, the general conclusion of Micheaux-Ferriere and Carron (1989) is in accordance with the results we have obtained for cyclamen and cucumber. The majority of cases of somatic embryogenesis described in literature use a callus phase on solid medium (Ammirato 1983, Zimmerman 1993). This seems to be the generally accepted method for obtaining somatic embryos, even when subsequently suspension cultures are made by dispersing the callus in liquid medium.It is our conviction that a callus phase should be avoided, since it is unnecessary laborious, and may lead to genetic instability and loss of embryogenic potential, as will be discussed below. Furthermore, with callus controlled growth is hard to establish, and in contrast to starting embryogenic cell lines in liquid medium, which is now possible, the large scale production of somatic embryos from callus cultures may not be feasible.Optimal proliferation of PEMs.Once embryogenic cell lines are obtained the PEMs have to be proliferated in order to produce the number of embryos wanted. For proliferation it is essential that PEMs are arrested in their development to embryos. Moreover, the number of PEMs must be multiplied by growth and subsequent disintegration. A single PEM, with a doubling time of 3.5 days, will produce 1 million PEMs in 10 weeks. The doubling times of plant cells, cultured in vitro , reported for cells of different species, range from 1.0 to 6.3 days, but it is not known whether embryogenic cell lines were used in all cases (Taticek et al. 1991). It will be obvious that the amounts of nutrients required for this proliferation are substantial. In our system, the proliferation was performed in flasks, and the cell lines were subcultured every 14 days. This included the complete replacement of the medium and dilution of the cells to a standard cell density per flask. Apart form the hormones, also the optimal nutrient composition of the maintenance medium differs for various crops. For cyclamen and carrot B5 medium was suitable, but for cucumber KK medium was used. For cucumber the medium composition had to be changed after the initiation phase. Initiation of cucumber cell lines was done in MS medium, but MS medium proved to be inadequate for the long term maintenance of the embryogenic cell line (Chapter 6).During a 14 day subculture period, nutrients are consumed by the growing cells. If the initial concentration of a nutrient in the medium is too low it may get depleted before the end of the subculture period, resulting in a decrease in growth rate. Indeed for cyclamen cell lines, growing in B5 medium, we observed depletion of ammonium after six days and this caused a decrease in growth rate (Chapter 5, Fig. 3). A similar observation was made for alfalfa growing in Schenk and Hildebrandt medium, containing 3 mM ammonium (McDonald and Jackman 1989). Uptake of nutrients were also measured in carrot cell lines growing in B5 medium, and it was found that phosphate could be depleted in 6 days (Ashihara and Nygaard 1989). Carrot cell lines even ceased growing, also known as the stationary phase, 10 days after subculturing due to sugar depletion (starting with 58 mM sucrose, Dijkema et al. 1988). Sugar depletion and subsequent cessation of growth were also observed in cucumber cell lines growing in MS medium with 86 mM of sucrose (Callebout and Motte 1988). Carrot cell lines showed an increased percentage of polyploid cells and loss of embryogenic potential, if subculture periods were extended and a stationary phase was reached (Bayliss 1975, Halperin 1966). Ashihara and Nygaard (1989) observed a decrease in RNA content of carrot cells during phosphate depletion. It was suggested that cells needed breakdown products of RNA, caused by reduced capacity for de novo synthesis. Depletions during the stationary phase may lead to impaired protein synthesis and affect control mechanisms, and may eventually lead to genetic instability (Bayliss 1975). Apparently it seems advantageous to avoid a stationary phase during a subculture period (Bayliss 1975).In the proliferation phase, it is essential to prevent the formation of embryos from PEMs. When this is not adequately dealt with, premature embryo formation will occur and all PEMs may be lost. The balance between the auxin and the cytokinin, and the concentration of the auxin, determine whether the PEMs will proliferate or develop into an embryo. In contrast to the initiation phase, where the hormones function as inducing compounds of the embryogenic potential, their role in the proliferation phase is totally different. The initial hormone concentration during a subculture period is important, and in general, a high auxin to cytokinin ratio is required. Low auxin to cytokinin ratios or low auxin concentrations may lead to premature embryo formation. We found that for cyclamen this ratio should be about 5 and for cucumber about 20. In cucumber, the increase in the 2,4-D concentration from 5 μM in the initiation phase to 45 μM in the proliferation phase, had an additional effect of a 6-fold increase in the number of PEMs per PCV (Chapter 6). In cyclamen such effect was however not observed (Chapter 5).In other studies on embryogenic cucumber cell lines (reviewed by Debeaujon and Branchard 1993), usually 2 to 5 μM of auxin is used. By comparison, the 45 μM 2,4-D used in our cucumber cell lines is very high. This difference is most probably explained by the rapid growth of our cell lines in KK medium. The actual growth rates of cucumber cell lines in other studies were never reported, but since in most cases callus cultures were used, the growth rates were probably much lower.The proliferation of embryogenic carrot cell lines in hormone-free medium containing ammonium as the sole nitrogen source, which results in low medium pH, was reported by Smith and Krikorian (1990). In that case PEMs were maintained and proliferated without the premature formation of embryos. This remarkable medium may be an example of the relationship between nutrients and hormones, noticed by Preece (1995), and illustrates how nutrients and hormones substitute for each other in specialized circumstances in carrot cell lines.An intriguing phenomenon was observed, in both cyclamen and cucumber cell lines. A large variation in the number of PEMs during a subculture period, in the size fractions 150- 300 μm and 100-150 μm, respectively, was observed (Chapter 5, Fig. 6, Chapter 6, Fig. 3). Due to growth of the cell lines, the number of PEMs increased, but the number of PEMs in a specific fraction varied during a subculture period. The specific fractions were chosen for their ability to produce single embryos, and represent only a small part of the whole biomass. The variation in the number of PEMs in these fractions may represent a small, but significant change in average PEM size during a subculture period. This might be related to the discontinuity of the process and a possible explanation is the following. At the start of the subculture period the nutrient and hormone concentrations in the medium are high, possibly resulting in a disintegration of large PEMs, and as a result, in an increase in the number of small PEMs (Halperin and Jensen 1967). Due to growth of the cells, the concentrations of nutrients and hormones in the medium decrease, possibly preventing PEMs from disintegrating, and consequently, decreasing the number of small PEMs in the specific fractions mentioned above. McDonald and Jackman (1989) measured that during a subculture period nutrients, pH of the medium, hormones and osmotic pressure varied dramatically, possibly explaining the change in the average size of the PEMs. Similar variations in aggregate size were observed for other species, but it was not mentioned whether embryogenic cell lines were used (Taticek et al. 1991).The maintenance of cell lines should be one of the best controlled phases, for the cell line is the source of all the embryos and has to be maintained for a prolonged period. PEMs have to be proliferated without loss of embryogenic potential and without premature embryo formation. We have shown that control of this phase can be achieved by medium adaptations and controlled subculture regimes.Production of somatic embryos.For producing somatic embryos, PEMs are sieved from a cell line, and then inoculated in hormone-free medium. In our studies on carrot, as well as cyclamen and cucumber, we observed that large PEMs tend to produce more than one embryo which are usually interconnected. On the other hand, very small PEMs tend to be less efficient in embryo formation, and a low percentage of these small PEMs actually forms an embryo. Chee and Cantliffe (1992) obtained similar results in sweetpotato, but their cell lines were produced from dispersed callus which may be the reason for the observed low efficiency of embryo formation.In order to produce as many single embryos as possible, and at the same time a high efficiency of embryo formation, for cyclamen and cucumber the best results with PEMs between 50 and 300 μm. The formation of embryos from PEMs, does not require exogenous hormones or other specific compounds, indicating that PEMs are fully capable of producing an embryo-like structure with as well a root and a shoot meristern. After transfer to hormonefree medium and dilution, the morphogenetic potential is 'released' and the embryo is formed, almost by itself. Successively, the PEMs develop into a globular, heart and torpedo shaped embryo. In cyclamen this is difficult to recognize due to the monocot-like nature of this dicot. It is our view that PEMs can be regarded as embryos arrested in their development, and the induction of embryo formation is therefore not induced in the hormone-free medium, but already during the initiation phase.The efficiency of embryo formation, expressed as the percentage of PEMs forming an embryo, can be high. In cucumber this usually was 10 to 20%, but in carrot and cyclamen it was more than 60%. Dijkema et al. (1988) showed that the efficiency of embryo formation of carrot cell lines, varied during a subculture period, with an optimum at 7 days. We have never observed such variation during a subculture period with cyclamen and cucumber. This might be related to the observed cessation of growth of the carrot cells at the end of the subculture period, under the conditions used by Dijkema et al. (1988). Suboptimal growth conditions during the proliferation phase apparently decreased the ability of PEMs to form embryos. It suggests that the quality of the PEMs is related to the growth circumstances in the proliferation phase.In our experience the nutrient composition required for optimal embryo development, may differ from that of the maintenance medium. With both cyclamen and cucumber the maintenance medium, B5 medium for cyclamen and KK medium for cucumber, had to be changed into MS medium for optimal embryo development. The low initial cell density during embryo development will undoubtedly determine the suitability of a medium. Apparently the developing embryo has nutrient demands different from the proliferating PEMs. For cucumber ABA has been reported to improve embryo morphology (Ladyman and Girard 1992), but ABA is not generally applied (Debeaujon and Branchard 1993). For cyclamen, a high sucrose content favoured the formation of embryos, and prevented the formation of only roots (Chapter 5, Fig. 8). High sucrose concentrations were also used by Wicart et al. (1984) in the formation of cyclamen embryos on callus. Specific additions, like hormones (ABA) or high sucrose concentrations, may be required to improve the efficiency or germination, but we found that, in principle, hormone-free medium and low cell densities are sufficient for embryo development.Genetic stability is essential.In Chapter 5 and 6 we showed that, when the generation of embryogenic cell lines was performed in liquid medium, genetic stable embryogenic cell lines were obtained, which could be maintained for years without loss of embryogenic potential.For the large scale propagation of plants, using in vitro techniques, genetic stability is an essential prerequisite, since plants that differ from the mother plant, have little value. In tissue culture, somaclonal variation is widespread (Bayliss 1980, Lee and Phillips 1988), and often found when 'undifferentiated' tissue, designated callus, is propagated on solidified medium. Whether callus is undifferentiated can be argued, but it is clear that growth of cell clumps on a solidified medium differs from growth of PEMs in liquid medium. Callus on agar plates often shows polyploidization (Ashmore and Shapcott 1989, Custers et al. 1990, Ezura and Oosawa 1994, Pijnacker et al. 1989, Schwenkel and Grunewald 1991). In Haplopappus gracilis it was demonstrated that callus cultures resulted in more polyploidization than suspension cultures (Ashmore and Shapcott 1989). In a callus clump, only a minor part of the cells is in contact with the medium and directly receives the nutrients required for growth. The majority of the cells will receive nutrients via the intercellular spaces or via cell-cell contact. The supply of nutrients will therefore be limited by diffusion and concentration gradients, and there may be shortages of essential nutrients. This may result in a lower growth rate, or even absence of exponential growth. The stress imposed on the cells in this way, may possibly result in polyploidization or other deviations in genetic constitution.Our observation that during the initiation of cucumber cell lines, growth of tetraploid cells is favoured above diploid cells, is remarkable. Starting from chimaeric explants, consisting of diploid and tetraploid cells, fully tetraploid cell lines could be obtained. No octaploid PEMs were detected, showing that polyploidization did not occur, and that the cell lines were genetically stable. The ploidy level of cell lines of a mixed ploidy level did not change after removing the explant during proliferation of the PEMs in the maintenance medium. As was shown in Chapter 6, the growth rate of diploid and tetraploid cells was therefore equal. PEMs are either diploid or tetraploid and fully diploid cucumber cell lines could be obtained by selection and further proliferation of individual PEMs (Chapter 6).In light, more fully tetraploid cell lines were formed than in the dark, and light therefore seemed to favour growth of tetraploid cells. However, cultures initiated in the dark could also show tetraploidization, indicating that light is not decisive. Since polyploidization was not observed in established embryogenic cucumber cell lines growing in the light, it seems that light only exerts its action on cells which are relatively undifferentiated, like cells of callus, or cells in the explant material de differentiating in liquid medium from which embryogenic cell lines are established. Indeed, initiation of carrot cell lines directly in liquid medium, in the light, if done according to De Vries et al. (1988), resulted in partial polyploid cell lines (Chapter 6). In accordance with this is the polyploidization observed in cucumber callus cultures during growth in the light, even when the cultures were initiated in the dark (Custers et al. 1990). Light during callus growth may also have caused the polyploidization reported for carrot (Coutos-Thevenot et al. 1990), melon (Ezura and Oosawa 1994) and cyclamen (Schwenkel and Grunewald 1988, 1991). It must be noted, that in these cases polyploidization in the cell cultures does not necessarily mean the tetraploidization of diploid cells, but can also have arisen from the faster growth of tetraploid cells, compared with diploid cells.We have found that, once fully diploid cell lines were obtained, genetic and embryogenic stability were ensured, provided the cells were maintained in a medium producing optimal growth. This proves that the use of liquid medium is superior to solidified media for the whole process of somatic embryogenesis, and makes clonal propagation without genetic instability possible.Factors influencing embryogenic potential.We further showed that, in addition to culture conditions and subculture regime, other compounds are able to influence the embryogenic potential of cell lines. In carrot cell lines, which had lost their potential to produce embryos, somatic embryogenesis was re-induced, if the cells were grown in the presence of arabinogalactan-proteins (AGPs) isolated from carrot seeds (Chapter 3). The same AGP preparation could increase the embryogenic potential during the proliferation of carrot PEMs. Similar results were recently obtained with Picea abies, in which the addition of seed AGPs to cell lines, previously unable to produce mature somatic embryos, resulted in a further maturation of somatic embryos (Egertsdotter and Von Arnold 1995).Different fractions of carrot seed AGPs, characterized by their binding to the monoclonal antibodies ZUM 15 and 18, had different activities (Chapter 4). The fractionated ZUM 18 AGPs had a hormone-like dose response curve, and increased the embryogenic potential of carrot cell lines at the very low concentration of about 2 nM. On the other hand, ZUM 15 AGPs decreased embryogenic potential. The activity of the ZUM 18 AGPs did not depend on the species from which they were isolated. Tomato ZUM 18 AGPs could also effect the embryogenic potential of carrot cell lines, and carrot ZUM 18 AGPs could influence the embryogenic potential of cyclamen cell lines. In contrast, unfractionated tomato seed AGPs had a low ability to increase embryogenic potential in carrot cell lines. The presence and ratio of different AGP epitopes therefore determined the overall effect of AGPs on the cells, and different epitopes appeared to have different activities. The different epitopes could be located on different, but also on the same molecule.The composition of the AGP mixture, excreted by carrot cell lines into the medium, changed as the cell lines got older (Chapter 3, Fig. 1), as demonstrated by the change in the crossed-electrophoresis patterns. Cell lines of different ages contain different AGPs and might therefore not respond in the same manner to subsequent, identical experimental conditions. When the AGPs of a cell line change this might result in a different response. It is our view that any experiment should be performed with cell lines of the same age, rather than with the same cell line over and over again.The conditions in the experiments, in which the activity of AGPs was demonstrated, were very different from the culture conditions used for embryogenic cell lines, as described in Chapter 5 and 6. In order to detect activity of AGPs, cells were grown at the lowest cell density allowing growth. Under these conditions competition between the added AGPs and the AGPs produced by the cells is avoided.The biological activity of AGPs remains intriguing. We showed that the development of cells, and even the expression of totipotency, can be altered by the addition of specific AGPs or AGP-epitopes. AGPs, derived from tissues containing embryogenic cells, i.e. the seeds, caused other cells to form somatic embryos. This indicates that AGPs may be involved in cellcell communication and/or cell identity. How AGPs exert their activity is still unknown, but specific epitopes seem to be involved. The assay system described in Chapter 3 and 4 and the monoclonal antibodies ZUM 15 and 18 might be used for further exploration of the role of AGPs and the different activities of different AGP-epitopes.Induction of embryogenic potential.Addition of specific AGPs to carrot cell lines can induce embryogenic potential, but some other AGPs can delay the onset of the induction (Chapter 3). Extracellular proteins excreted by embryogenic carrot cell lines are able to shorten the path towards an embryogenic carrot cell line as was demonstrated by De Vries et al. (11988). Various types of stress, like osmotic, salt and heavy metal treatment, are just as well able to induce embryogenic potential in carrot cell lines (Harada et al. 1990). Among the plant hormones, auxins are considered to be very effective in the induction of embryogenesis (Ammirato 1984, Zimmerman 1993). Also in our experience, the powerful auxin 2,4-D is able to induce cell division in certain parts of the explants, while in other parts, cells only elongate and the tissues disintegrate (Chapter 5 and 6). It seems that the onset of cell division in a specific set of cells, surrounding the vascular tissue, may lead to the formation of somatic embryos or PEMs. What causes these cells to divide and form embryos is not known, but their position in the explant and ability to react to a changed environment are likely important factors. It is our view that AGPs and extracellular proteins may have a function in this process.During the initiation phase of an embryogenic carrot cell line, meiotic-like cell divisions have been observed, leading to somatic embryos derived from segregants (Giorgetti et al. 1995). During the initiation phase, haploiclization of the cell line was observed, resulting in carrot cells containing 1 pg of DNA, being half of the 2 pg for diploid carrot cells reported by Bennet and Smith (1976). In contrast to this is the more recent finding of our and other laboratories, that diploid carrot cells contain 1 pg DNA (C. Kreuger, I Dijkstra, unpublished results and Bennet and Leitch 1995). Giorgetti et al. (1995) used Feulgen staining to quantify the DNA content, while we and Bennet and Leitch (1995) used flowcytometry. The discrepancy may be caused by the difficulty of staining carrot DNA with fluorescent dyes, although the cause of this is unknown. In flowcytometry more nuclei are measured and other dyes are used to stain the DNA, possibly resulting in more accurate determinations. As a consequence, the observed haploid cells by Giorgetti et al. would be diploid, originating from a tetraploid explant. Still, the segregation, as confirmed using RFLP and RAPD techniques, is not explained. Also in embryogenic callus of melon haploid cells were detected, starting from a diploid explant, but no relationship with the induction of embryogenic potential was made (Ezura and Oosawa 1994).Plants derived from segregants are not clones, and are therefore not desired. The cyclamen plants derived from somatic embryos, which were produced in our lab, do not show segregation at all. All plants are copies of the initial genotype, and any segregant would be easy to recognise. Meiotic cell divisions, if occurred at all, did therefore not contribute to the formation of the PEMs and embryos in cyclamen.Comparison of somatic embryogenesis with zygotic embryogenesis reveals many morphological similarities. However, the induction appears to be totally different. In the egg cell, fertilization starts the process of embryogenesis, no stress is involved al first glance. On the other hand, the change in the plant life cycle from sporophyte to gametophyte is accompanied by an extensive separation of the two generations (Bell 1992). In some cases plasmodesmata are no longer present for connection and cell to cell contact. In fact, the two generations might be regarded as opposing tissues. A similar isolation was observed during the formation of somatic embryos (Bell 1992). This isolation, either in response to a change of generations, or to an imposed stress, may cause cells to follow a specific developmental path, resulting in embryogenic potential and finally the formation of an embryo. In our view the formation of somatic embryos from the explant, is an answer to the stress, to escape the imposed conditions. Somatic embryogenesis may ensure the continuity of the individual, and may be a way to circumvent conditions lethal to the explant.The occurrence of meiotic cell divisions in explants or in establishing cell lines, may similarly be a response to the environmental conditions. However, embryos or plants derived from segregants have not been found in our or any other laboratory. This may be due to several reasons. Each laboratory has its own protocol for maintaining cell lines and a wide variation of media and culture regimes are applied, especially during the initiation phase. Different initiation protocols may also involve different ways of inducing embryogenic potential. The amount of stress used, or the types of stress may vary, possibly resulting in different responses of the explants. The formation of embryos by somatic cells might be achieved with or without meiotic cell divisions, depending on the inductive events during the initiation phase.Towards a large scale production of somatic embryos.The formation of somatic embryos by plants is, from a scientific point of view, a very attractive way to study the development of embryos, without the constraints of surrounding tissues. The other important aspect of somatic embryogenesis is its application in plant breeding. The cloning of superior plants may enhance breeding strategies and increase plant production. The research described in this thesis may contribute to both aspects of somatic embryogenesis. In the first part of the thesis it is shown that AGPs have a role in the embryogenic potential of cells. In the second part is described how correct initiation, directly in liquid medium, and controlled proliferation of PEMs lead to genetic and embryogenic stable cell lines. These conditions are essential for the large scale production of plants via somatic embryogenesis.