4399
Journal of Cell Science 113, 4399-4411 (2000) Printed in Great Britain © The Company of Biologists Limited 2000 JCS1834
Two waves of programmed cell death occur during formation and development of somatic embryos in the gymnosperm, Norway spruce Lada H. Filonova1, Peter V. Bozhkov1,*, Vladimir B. Brukhin1, Geoffrey Daniel2, Boris Zhivotovsky3 and Sara von Arnold1 1Department
of Forest Genetics, Uppsala Genetic Centre, Swedish University of Agricultural Sciences, Box 7027, S-75007 Uppsala, Sweden 2Department of Wood Science, Swedish University of Agricultural Sciences, Box 7008, S-75007 Uppsala, Sweden 3Institute of Environmental Medicine, Division of Toxicology, Karolinska Institutet, Box 210, S-17177 Stockholm, Sweden *Author for correspondence (e-mail:
[email protected])
Accepted 5 October; published on WWW 16 November 2000
SUMMARY In the animal life cycle, the earliest manifestations of programmed cell death (PCD) can already be seen during embryogenesis. The aim of this work was to determine if PCD is also involved in the elimination of certain cells during plant embryogenesis. We used a model system of Norway spruce somatic embryogenesis, which represents a multistep developmental pathway with two broad phases. The first phase is represented by proliferating proembryogenic masses (PEMs). The second phase encompasses development of somatic embryos, which arise from PEMs and proceed through the same sequence of stages as described for their zygotic counterparts. Here we demonstrate two successive waves of PCD, which are implicated in the transition from PEMs to somatic embryos and in correct embryonic pattern formation, respectively. The first wave of PCD is responsible for the degradation of PEMs when they give rise to somatic embryos. We show that PCD in PEM cells and embryo formation are closely interlinked processes, both stimulated upon withdrawal or partial depletion of
auxins and cytokinins. The second wave of PCD eliminates terminally differentiated embryo-suspensor cells during early embryogeny. During the dismantling phase of PCD, PEM and embryo-suspensor cells exhibit progressive autolysis, resulting in the formation of a large central vacuole. Autolytic degradation of the cytoplasm is accompanied by lobing and budding-like segmentation of the nucleus. Nuclear DNA undergoes fragmentation into both large fragments of about 50 kb and multiples of approximately 180 bp. The tonoplast rupture is delayed until lysis of the cytoplasm and organelles, including the nucleus, is almost complete. The protoplasm then disappears, leaving a cellular corpse represented by only the cell wall. This pathway of cell dismantling suggests overlapping of apoptotic and autophagic types of PCD during somatic embryogenesis in Norway spruce.
INTRODUCTION
extensive morphogenesis including cell movement and, in many cases, germline formation (Müller, 1997), in plants only the apical-basal plan, viz. developmental axis with shoot and root meristems, is established during embryogenesis. Most morphogenic processes in higher plants occur at the postembryonic stages, after seed germination (Goldberg et al., 1994); maintenance of a continuous germline was not established during plant evolution (Andrews, 1998). These observations point to a relatively simplified embryonic pattern formation in plants compared with animals. Consequently, one would expect a more limited occurrence of PCD in plant embryogenesis. To date, no systematic studies have been reported that utilize molecular and biochemical markers of PCD to investigate the implication of PCD in plant embryogenesis. Curiously, degradation of terminally differentiated suspensor cells, whose transient active role is promoting the growth of the embryo
During the last two decades, there has been intensive research on the mechanisms of programmed cell death (PCD). In the animal life cycle, the earliest manifestations of PCD can already be seen during embryogenesis, when some cells have to be sacrificed for the sake of correct embryonic pattern formation. Studies with PCD-defective embryos have shown that the embryonic pattern formation is disturbed and/or the animal dies prenatally as soon as the PCD pathway is artificially blocked through mutagenesis or targeted gene disruption (White et al., 1994; Kuida et al., 1996). There are two main functions that PCD serves in animal embryonic development: (1) sculpting the embryo, and (2) removal of unneeded structures (Jacobson et al., 1997; Song and Steller, 1999). Compared to animal embryogenesis, which involves
Key words: Plant embryogenesis, Norway spruce, Programmed cell death, Cell dismantling
4400 L. H. Filonova and others Fig. 1. A schematic representation of the SE PEM II developmental pathway of somatic embryogenesis in Norway spruce (adapted from Filonova et al., 2000; not PEM III drawn to scale). Proliferation of PEMs is AUXIN stimulated by auxin and cytokinin. An + no PGR individual PEM should pass through a CYTOKININ series of three characteristic stages (I, II and III) to transdifferentiate to somatic SE embryos (SE). At stage PEM I, a cell ABA PEM I aggregate is composed of a small compact clump of densely cytoplasmic cells adjacent to a single enlarged and vacuolated cell. Similar cell aggregates but possessing more than one vacuolated cell have been classified as PEM II. At stage PEM III, an enlarged clump of densely cytoplasmic cells appears loose rather than compact; polarity is disturbed. Withdrawal of plant growth regulators (PGRs) triggers embryo formation from PEM III, whereas abscisic acid (ABA) is necessary to promote further development of somatic embryos through late embryogeny to mature forms. Shown in red are the cells of PEMs and somatic embryos, which stain in situ red with acetocarmine. Blue colour corresponds to the cells of PEMs and somatic embryos, which stain in situ blue with Evan’s blue. Shown as dashed blue lines in the last but one stage of the pathway are the remnants of the degenerated suspensor in the beginning of late embryogeny.
proper at early stages of embryogenesis (Yeung and Meinke, 1993), is cited in modern literature as one of the typical examples of PCD in plant development (Jones and Dangl, 1996; Beers, 1997), although this issue was examined solely at the cytological level, in the mid 1970s to early 80s (Nagl, 1976; Nagl, 1977; Gärtner and Nagl, 1980), and has never been reexamined by employing more specific assays for PCD. One reason may be the difficulty in isolating a sufficient quantity of embryonic tissue at early stages of embryogenesis to be able to perform such assays. The early zygotic embryo is minute and is surrounded by endosperm in angiosperms and by megagametophyte in gymnosperms. Furthermore, even within the same plant, different ovules develop asynchronously (Owens, 1995), which hampers collecting embryos at a common developmental stage. The latter, however, seems to be an important prerequisite for studying PCD, which is likewise a highly asynchronous process by itself; cells within a population may begin PCD at very different times, and the length of the various phases of morphological change can vary from cell to cell (Collins et al., 1997; Messam and Pittman, 1998). Somatic embryogenesis represents one form of asexual plant development that can be induced in callus or somatic tissues. This provides the means for versatile control over embryo formation and development, and ultimately for isolation of sufficient quantities of developmentally homogenous tissues using molecular and morphological stage-specific markers. In this connection, we considered that somatic embryogenesis would be an attractive tool for studying PCD in plant embryogenesis. Previous studies in our laboratory defined the whole developmental pathway of somatic embryogenesis in the gymnosperm, Norway spruce (Pinaceae), through the use of the time-lapse tracking technique (Filonova et al., 2000). This pathway involves two broad phases, which in turn are divided into more specific developmental stages. The first phase is represented by proliferating proembryogenic masses (PEMs), cell aggregates that can pass through a series of three characteristic stages distinguished by cellular organization and cell number (stages PEM I, II and III; Fig. 1), but can never develop directly into a real embryo. The second phase encompasses development of somatic embryos. The latter arise de novo from PEM III,
and then proceed through the same, stereotyped sequence of stages as described for zygotic embryogeny of Pinaceae (Singh, 1978; Fig. 1). Plant growth regulators (PGRs), auxins and cytokinins, are necessary during the first phase to maintain PEM proliferation, whereas embryo formation from PEM III is triggered by the withdrawal of PGRs. Once early somatic embryos have formed, their further development to mature forms requires abscisic acid (ABA; Fig. 1). We have found both direct and indirect indications that cell death is an important component of the described developmental pathway (Filonova et al., 2000). First, PEMs have been shown to be composed of two types of cell: small densely cytoplasmic cells, which stain red with acetocarmine, and enlarged, highly vacuolated cells, which are more or less elongated and permeable to Evan’s blue (Fig. 1). The latter, being an isomer of the more commonly used dye Trypan blue, can also be used to distinguish viable from dead or dying cells, based on its inability to pass through an intact plasma membrane (Huang et al., 1986). Secondly, a proper timing of suspensor degradation (within a few days after addition of ABA) has been suggested to be an important factor for normal embryo development. For convenience, we use the term ‘suspensor’ (or ‘embryo-suspensor’) in this paper to denote a massive structure that is formed during early zygotic and somatic embryogeny of gymnosperms and classified, in classic embryology, as secondary suspensor (Singh, 1978; von Aderkas et al., 1991). Thirdly, we have found differential expression of monoclonal antibody (mAb) JIM13-reactive epitope of arabinogalactan proteins (AGPs) in the cell walls of PEMs and somatic embryos. The vast majority of PEM cells expressed this epitope, but none of the cells in the early somatic embryos (Filonova et al., 2000). There is increasing evidence that AGPs are implicated in PCD in plants (Gao and Showalter, 1999; Majewska-Sawka and Nothnagel, 2000), and that it is specifically the JIM13-reactive epitope of AGPs which marks the cells destined for PCD (Schindler et al., 1995; Stacey et al., 1995). In this context, the obvious question is whether PEM cells of Norway spruce are committed to die. The present study tries to answer to this question, and also shows how the frequency of PCD in PEMs changes in the course of sequential treatments that regulate transition from PEMs (the first phase) to embryos (the second phase).
Programmed cell death in plant embryogenesis 4401 Furthermore we have examined a spatial distribution of dying cells in PEMs and in developing embryos. Finally, we suggest the cytological pathway of cell dismantling during somatic embryogenesis of Norway spruce. MATERIALS AND METHODS Cell culture of Norway spruce Embryogenic cell line 95.88.17 of Norway spruce (Picea abies L. Karst.), which has previously been used in time-lapse tracking (Filonova et al., 2000), was also chosen for the present study. Cell aggregates (i.e. a mixture of PEMs and somatic embryos) were stored in liquid nitrogen and thawed 5 months prior to the onset of experiments (Norgaard et al., 1993). Culture regrowth was achieved by inoculating cell aggregates on solidified half-strength LP proliferation medium (modified after Bozhkov and von Arnold, 1998) containing 9.0 µM auxin, 2,4-dichlorophenoxyacetic acid (2,4-D), and 4.4 µM cytokinin, N6-benzyladenine (BA). Thereafter, 2 months prior to the onset of experiments, proliferating cell aggregates were transferred to the liquid half-strength LP proliferation medium. A stable proliferating embryogenic suspension culture was established by weekly subculturing 3 ml of settled cell aggregates into 47 ml fresh proliferation medium in 250-ml Erlenmeyer flasks. This routine minimised the risk of occurrence of intraclonal variation in culture morphology among different flasks. Suspension cultures were grown on a gyratory shaker (100 rpm) in darkness at 22±1°C. In order ‘to switch’ the culture from proliferation to development (i.e. to induce somatic embryo formation from PEMs), auxin and cytokinin should be withdrawn (Filonova et al., 2000; Fig. 1). This was accomplished through two successive washings of 3-ml settled cell aggregates with 10-ml samples of modified full-strength BMI-S1 PGRfree medium (Bozhkov and von Arnold, 1998) in 15-ml Falcon tubes followed by inoculation of 3-ml supernatant-free cell aggregates into 47 ml BMI-S1 medium of the same composition in 250-ml Erlenmeyer flasks. The cultures were grown on a gyratory shaker (100 rpm) in darkness at 22±1°C for 1 week until subjected to ABA treatment enabling further somatic embryo development to mature forms. ABA treatment involved plating 1 ml of settled cell aggregates onto Whatman no. 2 filter paper (5.5 cm in diameter) placed on solidified BMI-S1 maturation medium with 30 µM ABA in 60-mm Petri plates (10 ml medium per plate; Bozhkov and von Arnold, 1998). The filter papers with developing somatic embryos were transferred to fresh medium after 2, 4 and 6 weeks. The cultures were incubated in the dark at 22±1°C. Sampling of cell cultures Cell cultures were sampled for subsequent analyses (see following sections) at four sequential time points, with a progressive increase in the frequency of somatic embryos compared to PEMs (Table 1): (1) 3 days after subculture in auxin- and cytokinin-containing proliferation medium (designated as 3d 2,4-D+BA), (2) at the end of subculture in proliferation medium (designated as 7d 2,4-D+BA), (3) at the end of subculture in PGR-free medium (designated as 7d no PGR) and (4) 7 days after plating on ABA-containing medium (designated as 7d ABA). In addition, individual somatic embryos at advanced stages of development were collected after 2 and 7 weeks of ABA treatment for terminal deoxynucleotidyl transferase (TdT)mediated dUTP nick-end labeling (TUNEL) assay. Human apoptotic cells Jurkat cells (human leukemia T cell line, obtained from the European Collection of Cell Cultures, Salisbury, UK) were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum, 2 mM glutamine, 100 i.u./ml penicillin and 100 mg/ml streptomycin in a humidified atmosphere of 5% CO2 in air at 37°C. Cells were
Table 1. Composition of embryogenic cell line of Norway spruce at four sequential sampling times Composition, % Time point 3d 2,4-D+BA 7d 2,4-D+BA 7d no PGR 7d ABA
PEM I
PEM II
PEM III
SE
26 8 0 0
36 26 6 0
32 47 19 1
6 19 75 99
The percentage composition of PEMs (at stages I, II and III) and early somatic embryos (SE) were determined after observation of at least 500 individual cell aggregates with at least four replicates on every occasion. d, days.
maintained in logarithmic growth phase by routine passage every 3-4 days. Apoptosis was induced by treatment with 250 ng/ml agonistic anti-CD95 mAb (clone CH-11, Medical & Biological Laboratories, Ltd., Nagoya, Japan). 3 hours after incubation, cells were collected by centrifugation at 1000 g for 5 minutes and the pellet resuspended in phosphate-buffered saline (PBS) and subjected to a second centrifugation. The supernatant was aspirated and the pellet kept at −70°C before DNA was extracted. The fragmented DNA of Jurkat cells served as a positive control for Norway spruce DNA in electrophoresis experiments. Protoplast isolation and fractionation Protoplasts were isolated at four sequential time points (Table 1). The cell wall digesting enzyme solution consisted of 1% (w/v) Cellulase ‘Onozuka R-10’ (Duchefa, Haarlem, The Netherlands), 0.25% (w/v) Macerozyme R-10 (Duchefa, Haarlem, The Netherlands), 0.25% (w/v) Driselase (Sigma, St Louis, MO, USA), 5 mM CaCl2 and 0.4 M mannitol, pH 5.8. With suspension cultures (i.e. the first three time points), cell aggregates were first collected on the top of an 80-µm nylon screen and then transferred to 90-mm Petri plates containing the enzyme solution (12 ml for 2 g fresh mass of cell aggregates), whereas 7d ABA-cell aggregates were placed directly in the enzyme solution and then resuspended by using a transfer pipette. Cell aggregates were incubated in the enzyme solution for 1.5 hours on a gyratory shaker (50 rpm) at 25±1°C in the dark. It is significant that under the specified conditions, only PEM cells released protoplasts, as determined by monitoring cell aggregates incubated in the enzyme solution at 10-minute intervals using an inverted microscope (Axiovert 10, Zeiss, Germany). Cells within the compact embryonal masses of somatic embryos remained intact due to the presence of the surrounding protoderm layer, whereas long suspensor cells, on the contrary, collapsed in the presence of hydrolytic enzymes. As the proportion of PEMs to somatic embryos dramatically descreased from 3d 2,4-D+BA to 7d ABA (Table 1), the fresh mass of the culture sample was correspondingly increased at each succeeding time point to yield a sufficient number of protoplasts. Following enzyme treatment, the mixture of protoplasts, embryonal masses and cell wall debris was filtered through a series of nylon screens with successive 400, 200, 150, 80, 50 and 30 µm pore sizes. The fraction of large protoplasts separated on the top of the 30 µm screen was then resuspended in 10 ml of washing solution containing 5 mM CaCl2 and 0.4 M mannitol (pH 5.8) in 15-ml Falcon tubes, and was centrifuged at 80 g for 5 minutes. The supernatant was removed and the pelleted protoplasts were subsequently washed by the same procedure twice more in order to remove nucleases (if any were present), recently suggested to be contaminating other types of Onozuka cellulase, produced by Yakult Honsha Co. Ltd., Tokyo, Japan (Bethke et al., 1999). The mixture of small protoplasts and cell debris passing through the 30 µm screen was centrifuged at 80 g for 5 minutes, the supernatant removed, and the pellet subjected to the above-described three-step
4402 L. H. Filonova and others washing procedure for removing putative nucleases. The pellet was then resuspended in washing solution (9 ml washing solution added to 1 ml pellet), and gently mixed with an equal volume of 30% (w/v) Ficoll 400 (Amersham Pharmacia Biotech AB, Uppsala, Sweden) in washing solution. 2 ml of this mixture was layered over 2 ml of 20% Ficoll in 15-ml Falcon tubes and overlaid with 2 ml of 10% Ficoll (both in washing solution), and then with 0.5 ml of washing solution containing no Ficoll. The resulting discontinuous Ficoll gradient was centrifuged at 100 g for 20 minutes. Intact small protoplasts were collected from the two top interfaces with a Pasteur pipette, pooled and washed three times as previously described, to remove Ficoll. The time-lapse tracking performed during the protoplast isolation procedure has shown that small protoplasts were predominantly released from PEM cells, which stain red with acetocarmine, while large protoplasts originated from Evan’s blue-positive cells of PEMs. The viability of the protoplasts was assessed using fluorescein diacetate (FDA, Sigma, St Louis, MO, USA) (Widholm, 1972). Samples of small or large protoplasts were incubated for 5 minutes in FDA at 10 µg/ml and mounted on Polysine slides (Menzel-Gläser, Germany). Over 500 protoplasts with at least three replicates were scored on every occasion with a Microphot-FXA (Nikon, Japan) fluorescence microscope, equipped with a super high pressure mercury lamp supply (HB-10101AF) and the standard set of filters for fluorescein (B-2A). Both the living protoplasts showing yellowgreen fluorescence and the percentage of protoplasts lacking viability and excluding FDA (i.e. FDA-negative) were counted. Samples were taken from the large and small protoplast fractions for electrophoretic DNA fragmentation analyses and TUNEL assay. In a separate series of experiments, protoplasts isolated from PEMs at 3d and 7d 2,4-D+BA and 7d no PGR were not fractionated but directly subjected to TUNEL assay. Electrophoretic DNA fragmentation analyses In order to resolve large DNA fragments, the whole cell culture or fractionated protoplasts were subjected to pulsed-field gel electrophoresis (PFGE). Samples of the whole cell culture (Table 1) or protoplasts were mixed with an equal volume of molten 2% (w/v) low-melt preparative grade agarose (Bio-Rad, CA, USA) in 0.5× TBE buffer (Sambrook et al., 1989) and transferred to a mold at 4°C. Solidified agarose blocks were transferred into a lysis buffer containing 10 mM Tris-HCl (pH 7.6), 100 mM EDTA, 20 mM NaCl, 1% (w/v) Sarcosyl and 5 mg/ml proteinase K (Boehringer Mannheim), and incubated at 37°C for 40 hours. Electrophoresis was carried out in 1% (w/v) pulsed field certified agarose (Bio-Rad, CA, USA)-gel using CHEF-DR II apparatus (Bio-Rad). The gels were run at 5.7 V/cm in 0.5× TBE buffer with pulse ramping time from 10 to 60 seconds at 14°C for 24 hours. Agarose gels were stained with 1 µg/ml ethidium bromide for 30 minutes and destained with distilled water for 3 hours. For size determination, PFGE Marker I (Boehringer Mannheim) was used as standards. For assessment of internucleosomal fragmentation, DNA from the whole cell culture, fractionated protoplasts and apoptotic Jurkat cells was isolated using the proteinase K-based method (Allen and Newland, 1998). In the case of whole cell cultures, the cell walls of intact cell aggregates were broken down by a short (30 seconds) destructive pretreatment in 2 ml FastRNA Tubes-Green (BIO 101, Inc., Vista, CA, USA) using Fast Prep FP120 apparatus (Savant Instruments, Inc., Farmingdale, NY, USA) prior to DNA isolation. For each sample, 4 µg DNA was loaded per lane and electrophoresed in 1.8% MetaPhore agarose (FMC BioProducts, Rockland, ME, USA)-gel at 6 V/cm using 0.5× TBE buffer. The gels were run for 3 hours, stained with 1 µg/ml ethidium bromide for 30 minutes and destained with distilled water for further 3 hours. For size determination, 100 bp DNA Ladder (MBI Fermentas, Vilnius, Lithuania) was used as standards. In situ detection of DNA fragmentation (TUNEL assay) Norway spruce protoplasts, whole mounts of PEMs and early somatic
embryos, as well as sections of somatic embryos at the beginning and end of late embryogeny, were all subjected to TUNEL assay (Gavrieli et al., 1992), using the fluorescein-dUTP-based in situ death detection kit (Boehringer Mannheim). Fractions of small and large protoplasts, as well as non-fractionated protoplasts, were fixed by mixing with an equal volume of 4% (w/v) paraformaldehyde in PBS (pH 7.2), dropped on Polysine slides and dried for 1 hour at 40°C. The preparations of whole-mount PEMs (sampled at 3d and 7d 2,4-D+BA) and early somatic embryos (sampled at 7d no PGR and 7d ABA) were fixed in 4% (w/v) paraformaldehyde in PBS (pH 7.2) for 1 hour under vacuum at 25±1°C. Somatic embryos at the beginning and end of late embryogeny (sampled after 2 and 7 weeks of ABA treatment, respectively) were sequentially fixed, dehydrated and embedded as previously described (Filonova et al., 2000), the only difference being that JB-4TM Embedding Kit (Polyscience, Inc., Warrington, CA, USA) instead of Technovit resin was used in the present study. The embedded embryos were sectioned on a motorised microtome (HM 350, Microm, Germany), and the sections (3.5 µm) were placed on Polysine slides. The fixed protoplasts were labeled with TUNEL kit diluted 1:2 in reaction buffer (Boehringer Mannheim) (Danon and Gallois, 1998), excluding permeabilization with proteinase K. The whole mounts and sections were labeled with TUNEL kit according to the manufacturer’s protocol, without dilution (Boehringer Mannheim). Negative controls were included by omitting TdT. For nuclear staining, the samples labeled with TUNEL kit were washed twice by PBS and stained with 1 µg/ml 4′-6-diamino-2phenylindole (DAPI, Boehringer Mannheim) in S-buffer (Kuroiwa et al., 1992). Finally all samples were washed with distilled water, covered by Fluorsave (Calbiochem, CA, USA) and examined with a Microphot-FXA fluorescence microscope using FITC and UV-2A sets of filters for TUNEL detection and DAPI, respectively. Kodak Elite 400 film was used for photomicrography. The percentage of TUNEL-positive protoplasts was determined by scoring over 500 protoplasts with at least three replicates at every occasion. Prior to TUNEL assay, individual PEMs at stages I, II and III, as well as early somatic embryos, were preselected using an inverted microscope. An assessment of the distribution of TUNELpositive cells within PEMs and somatic embryos was based on the analysis of least 20 whole-mount or sectioned PEMs or somatic embryos at a certain developmental stage. Electron microscopy For ultrastructural analysis, PEMs III and somatic embryos collected at 7d 2,4-D+BA and 7d no PGR, respectively, were first embedded in thin layers of 0.6% (w/v) Seaplaque agarose (FMC BioProducts, Rockland, ME, USA) dissolved in distilled water (pH 5.8). Preselected cell aggregates were cut from agarose layers leaving enough agarose around to protect cells from injury during fixation and embedding. Fixation and dehydration were carried out according to Fowke (Fowke, 1995). Samples were embedded in LR-White Hard Grade resin (Ted Pella Inc., Redding, CA, USA) using an infiltration protocol previously described for Technovit resin (Filonova et al., 2000). Cell aggregates were sectioned using a Reichert Ultracut E/FC4D (Leica, Germany) ultramicrotome and sections were collected on formvar coated copper grids. Sections were observed unstained using a Philips CM/12 transmission electron microscope operated at 60 kV. Observations were recorded using Kodak 4489 film.
RESULTS Somatic embryo formation is accompanied by a massive DNA fragmentation in PEM cells In embryogenic cell cultures of Norway spruce, withdrawal of PGRs leads to a profound alteration in the balance between
Programmed cell death in plant embryogenesis 4403 PEMs and early somatic embryos through stimulating somatic embryo formation and correspondingly suppressing PEM multiplication (7d 2,4-D+BA cf. 7d no PGR; Table 1). Even a partial depletion of auxin and cytokinin towards the end of subculture in the proliferation medium results in an appreciable increase in the frequency of somatic embryos (3d cf. 7d 2,4D+BA; Table 1). Treatment with ABA promotes continuous embryo development, with a complete disintegration of the few remaining PEM III (7d ABA; Table 1). The integrity of DNA in the whole cell culture sampled at four sequential time points was assessed by PFGE and conventional agarose gel electrophoresis. Two levels of DNA fragmentation were observed in all samples. First, the presence of large fragments of about 50 kb and under, known to be a marker for apoptosis (Bortner et al., 1995), was detected by PFGE (Fig. 2A). Secondly, when subjected to conventional agarose gel electrophoresis, isolated DNA from Norway spruce revealed an oligonucleosomal ladder, similar to that of human apoptotic cells (Fig. 2B), indicating fragmentation of DNA into multiples of approximately 180 bp. The occurrence of DNA fragmentation in the whole culture consisting of a number of cell types aggregated in PEMs and somatic embryos did not, however, suggest how the frequency of cells with fragmented DNA changes in the course of somatic embryogenesis and whether this frequency correlates with somatic embryo formation. The differential responses of PEM and embryo cells to hydrolytic enzymes made it possible to isolate intact protoplasts only from PEMs and not from embryos (see Materials and Methods) for the quantitative assessment of DNA fragmentation in PEMs with TUNEL. Isolated protoplasts demonstrated a moderate increase in the frequency of TUNEL-positive nuclei, from 19% to 24%, upon the transition from 3d to 7d 2,4-D+BA, followed by a pronounced increase to 50% after withdrawal of PGRs (7d no PGR; Fig. 3). Interestingly, regression analysis has shown very strong positive correlation between the frequency of somatic embryo formation and the percentage of PEM-protoplasts fragmenting DNA (r=0.99, P30 µm) and small (
