Re-organisation of the cytoskeleton during developmental programmed cell death in Picea abies embryos



Cell and tissue patterning in plant embryo development is well documented. Moreover, it has recently been shown that successful embryogenesis is reliant on programmed cell death (PCD). The cytoskeleton governs cell morphogenesis. However, surprisingly little is known about the role of the cytoskeleton in plant embryogenesis and associated PCD. We have used the gymnosperm, Picea abies, somatic embryogenesis model system to address this question. Formation of the apical–basal embryonic pattern in P. abies proceeds through the establishment of three major cell types: the meristematic cells of the embryonal mass on one pole and the terminally differentiated suspensor cells on the other, separated by the embryonal tube cells. The organisation of microtubules and F-actin changes successively from the embryonal mass towards the distal end of the embryo suspensor. The microtubule arrays appear normal in the embryonal mass cells, but the microtubule network is partially disorganised in the embryonal tube cells and the microtubules disrupted in the suspensor cells. In the same embryos, the microtubule-associated protein, MAP-65, is bound only to organised microtubules. In contrast, in a developmentally arrested cell line, which is incapable of normal embryonic pattern formation, MAP-65 does not bind the cortical microtubules and we suggest that this is a criterion for proembryogenic masses (PEMs) to passage into early embryogeny. In embryos, the organisation of F-actin gradually changes from a fine network in the embryonal mass cells to thick cables in the suspensor cells in which the microtubule network is completely degraded. F-actin de-polymerisation drugs abolish normal embryonic pattern formation and associated PCD in the suspensor, strongly suggesting that the actin network is vital in this PCD pathway.




microtubule-associated protein


proembryogenic mass


plant growth regulator


programmed cell death


terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labelling


abscisic acid


Embryogenesis in planta takes place under multiple layers of tissues so that its direct observation is obscured. In order to study the mechanisms of plant embryo development, it has been necessary to develop model systems for in vitro embryogenesis. Carrot (Daucus carota) somatic embryogenesis has long been known as the most advanced system to study embryo development in angiosperms (Mordhorst et al., 1997). We have developed a similar model system of somatic embryogenesis for the gymnosperm, Norway spruce (Picea abies; for review see Von Arnold et al., 2002).

This system includes a stereotyped sequence of developmental stages and corresponding regulatory treatments (Bozhkov et al., 2002; Filonova et al., 2000a,b; Figure 1a). Plant growth regulators (PGRs), auxin and cytokinin, stimulate proliferation of proembryogenic masses (PEMs) and, in contrast, suppress embryo formation. Consistently, withdrawal of PGRs triggers the PEM-to-embryo transition with concomitant activation of the PEM degradation programme. Early somatic embryos require abscisic acid (ABA) for their further development and maturation. The development of embryos therefore can be synchronised by specific PGR treatments. It is noteworthy that following the PEM-to-embryo transition, the development of P. abies somatic embryos is similar to the basal plan of zygotic embryogeny in the Pinaceae (Singh, 1978). However, some genotypes deviate from normal embryo pattern formation exhibiting developmental arrest at certain stages (Filonova et al., 2000b).

Figure 1.

Development of normal (88.17) and arrested (88.1) embryogenic cell lines of Picea abies.

Cell line 88.17 is capable of embryo development and has a large proportion of cells undergoing PCD, whereas cell line 88.1 is arrested at the PEM-to-embryo transition and has a small proportion of dying cells.

(a) Developmental pathway of somatic embryogenesis in P. abies and the stages analysed in this study. Cell line 88.17 successfully passes through the PEM-to-embryo transition (shown by the bold arrow) and then gradually develops into mature embryos. In contrast, cell line 88.1 is arrested at the PEM-to-embryo transition (block shown as a solid square) and remains at the PEM stage. Shown in red are the cells with dense cytoplasm, which give red colouration with acetocarmine. The blue represents the vacuolated cells of PEMs and embryos which undergo PCD and stain blue with Evan's blue (Filonova et al., 2000a).

(b) Treatments and sampling time points on the same scale as (a). In progressively developing cell line 88.17, five sampling time points corresponded to five major developmental stages. Sample 1 taken 3 days after subculture into PGR-containing medium contained predominantly small PEMs classified as PEM I and PEM II. Sample 2 collected 7 days after subculture into PGR-containing medium contained large PEMs classified as PEM III. Sample 3 was taken 24 h after withdrawal of PGRs when large scale PEM-to-embryo transition occurs. Sample 4 collected 7 days after withdrawal of PGRs was mainly composed of early somatic embryos. The last sample taken 7 days after addition of ABA represented exclusively somatic embryos in the beginning of late embryogeny. In contrast, none of the samples collected in cell line 88.1 contained somatic embryos, as this line is arrested at the PEM-to-embryo transition.

(c) Percentage composition of PEMs and embryos at five sequential sampling time points in both cell lines. Note the invariant composition of the developmentally arrested cell line 88.1 (containing PEMs only) as opposed to cell line 88.17 showing increased frequency of somatic embryos and concomitant elimination of PEMs.

(d) Frequency of TUNEL positive nuclei in PEMs collected at the first four sampling time points in both cell lines. Note the very low frequency of PCD in cell line 88.1 as compared to cell line 88.17.

The important aspect of early embryogeny in P. abies is polarisation of the embryo via the formation of two distinct structures: the embryonal mass and the suspensor, which are linked by the embryonal tube cells. The embryonal mass cells are small and spherical with dense cytoplasm and mitotic activity. The asymmetric divisions of the most basally situated cells within the embryonal mass give rise to a layer of elongated embryonal tube cells which differentiate to form one layer of the suspensor cells. By this means, reiterated asymmetric divisions in the embryonal mass are continuously adding new layers of cells to the growing embryo suspensor which becomes composed of several layers of highly vacuolated elongated cells by the end of early embryogeny (Filonova et al., 2000b; Singh, 1978).

We have shown that the embryo suspensor is a terminally differentiated structure that is eliminated through programmed cell death (PCD) at the end of early embryogeny (Filonova et al., 2000a, 2002). Cytoplasmic degradation features major signs of autophagy, whilst nuclear disassembly proceeds through the pathway characteristic of apoptosis. The earliest morphological sign of PCD in the embryo suspensor is an increase in the number and activity of Golgi complexes. Then, autolytic vacuoles and vesicles accumulate in the cytoplasm and form one or more large vacuoles that take up the majority of the cell volume. Nuclear DNA fragments and, finally, the vacuole collapses leaving a hollow-walled corpse (Filonova et al., 2000a).

Bearing in mind the results of these previous observations on embryonic pattern formation and associated PCD in P. abies, we speculated that an early somatic embryo should possess a gradient of cells at different stages of PCD along its apical–basal axis, starting with living meristematic cells in the embryonal mass to cell corpses at the distal end of the embryo suspensor. The transition from living cells to cell corpses is accompanied by changes in cell shape and cytoplasmic organisation (Filonova et al., 2000a,b). It is well known that the architecture of cytoskeleton changes dramatically in the course of PCD in animals (Bursch, 2001; Bursch et al., 2000; Jochova et al., 1997; Srivastava et al., 1998). However, the mechanisms of cytoskeletal re-organisation and regulation during plant embryo development and PCD remain unknown.

In this paper we have examined the re-organisation of the cytoskeleton in the three cell types that constitute the apical–basal embryonic pattern in P. abies. In particular, we ask whether the microtubule network is re-organised and in which cell types, and in light of these observations discuss a possible role for the plant microtubule-associated protein, MAP-65 (Jiang and Sonobe, 1993; Lloyd and Hussey, 2001; Smertenko et al., 2000), in the commitment to embryogenesis. Moreover, we study the role of F-actin in embryonic pattern formation with the aid of drugs that disrupt the actin network and discuss whether this network is essential for the execution of embryogenesis and PCD.

Results and discussion

Developmental pathway of somatic embryogenesis in Picea abies

The normal developmental pathway of P. abies somatic embryogenesis involves two major phases: proliferation of PEMs and development of the somatic embryos (Figure 1a). Proembryogenic masses can pass through a series of three characteristic stages, PEM I, PEM II and PEM III, that are distinguished by shape and cell number but can never develop directly into normal embryos (Filonova et al., 2000b). The latter arise de novo from PEM III and proceed through the same stages of early and late embryogeny as zygotic embryos of Pinaceae (Singh, 1978). Auxin and cytokinin support PEM proliferation, whereas the PEM-to-embryo transition is triggered by the withdrawal of these PGRs (Figure 1b) leading to a dramatic increase in the frequency of somatic embryos and concurrent PEM degradation in the whole cell culture. Contrary to early embryogeny, which does not require exogenous hormones, late embryogeny is promoted by ABA treatment (Figure 1a,b).

The somatic embryogenesis pathway of cell line 88.17 used in the present study corresponds well with this model. We will first discuss changes in the cytoskeleton associated with PCD during differentiation of somatic embryos in cell line 88.17. Thereafter, we will compare the organisation of the cytoskeleton in this normal cell line with that in a PCD-deficient and developmentally arrested cell line 88.1 (Figure 1). Sequential sampling of line 88.17 at five critical time points showed a progressive increase in the proportion of embryos versus PEMs, with the most pronounced increase in embryo frequency induced by withdrawal of PGRs (Figure 1c). The frequency of nuclei with fragmented DNA in PEMs was analysed at the first four sampling time points using terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labelling (TUNEL). The number of TUNEL positive cells in PEMs increased twofold, and at the same time the frequency of PEMs decreased from 66 to 25% (compare sampling points 3 and 4 in line 88.17 in Figure 1c,d). This strongly suggests that PEMs are eliminated by PCD during the PEM-to-embryo transition.

Organisation of the cytoskeleton during embryonic pattern formation and PCD

A typical early embryo from cell line 88.17 is shown in Figure 2. The cells in the embryonal mass are brightly imaged with phase contrast optics due to the high cytoplasmic content (Figure 2a); these cells are small and globular. Embryonal tube and suspensor cells on the other hand have large vacuoles and give a dim image using these phase contrast optics (Figure 2a); these cells are large and mostly elongated. Staining the nuclei with 4′-6-diamino-2-phenylindole (DAPI) shows the high density of nuclei in the embryonal mass compared to the lower density in the suspensor region (Figure 2b). The cells at the late stages of PCD showing TUNEL positive nuclei are located at the distal end of the embryo suspensor (Figure 2c). We have previously shown that nuclear degradation during embryonic PCD in P. abies involves dismantling of the nuclear pore complex, the lobing-like segmentation of nuclei and the release of chromatin loop-sized 50-kb DNA fragments followed by internucleosomal fragmentation (Filonova et al., 2000a).

Figure 2.

Organisation of microtubules in cell line 88.17.

An early embryo isolated from cell line 88.17 at sampling time point 4 showing its composite different cell types.

(a) Phase contrast image of the embryo. Inset is a schematic drawing of the embryo. EM, embryonal mass; ET, embryonal tubes; ES, embryo suspensor.

(b) DAPI staining of the same embryo.

(c) TUNEL staining of the same embryo. The TUNEL positive nuclei are located at the distal end of the suspensor.

(d) Staining of the microtubules in the same embryo with anti-α-tubulin antibodies. The scale bar is the same for a–d and equals 50 μm. The insets show higher magnification of the embryonal mass region (section 1), embryonal tube (section 2) and embryo suspensor cells (section 3). Scale bar corresponds to 10 μm.

(e) Staining of microtubules with anti-α-tubulin antibodies (shown in green) and fragmented nuclear DNA (TUNEL positive nucleus is shown in red) in the distal end of the suspensor. Scale bar corresponds to 10 μm.

Microtubule organisation progressively changes along the apical–basal axis of the embryo (Figure 2d, sections 1–3). The meristematic cells within the embryonal mass contain a fine microtubule network during interphase (Figure 2d, section 1); mitotic microtubule arrays are also observed in this part of the embryo (not shown). In embryonal tube cells, two orders of microtubule organisation could be identified: a transverse cortical microtubule network in some of the cells which is consistent with the cells’ elongate shape, and an array of disorganised microtubule fragments and aggregates in others (Figure 2d, section 2). In all the suspensor cells, only microtubule fragments and tubulin aggregates could be found in the cytoplasm (Figure 2d, section 3). Finally, in TUNEL positive cells no microtubule fragments were observed (Figure 2e).

These data show that the magnitude of microtubule disorganisation during early embryogeny of P. abies is increased along the embryo apical–basal axis. Indeed, the disorganisation of the microtubule network is already initiated after the asymmetric division of the embryonal mass cells, i.e. in the daughter embryonal tube cells which represent the earliest stage of suspensor differentiation. The microtubules undergo complete degradation in highly vacuolated suspensor cells which is followed by DNA fragmentation and vacuole collapse (Filonova et al., 2000a).

Though microtubule disorganisation is seen in the embryonal tube cells, the latter still develop into elongate cells to form the embryo suspensor. This suggests that there is a structure other than microtubules that controls the organisation of the cytoplasm and the shape of the cells in the suspensor. Analysis of the actin network in the embryos shows that this remains in the cells that lack the microtubule network (Figure 3). In the embryonal mass cells, a network of F-actin is observed (Figure 3b), and in the embryonal tube and suspensor cells it is re-organised into longitudinal and oblique cables (Figure 3c,d). Therefore, it is reasonable to suggest that actin supports the differentiation of the suspensor cells and controls cytoplasmic architecture in the absence of microtubules.

Figure 3.

Visualisation of actin filaments in an early embryo isolated at sampling time point 4 from cell line 88.17 using anti-actin monoclonal antibody clone C4.

(a) Whole embryo image. This image represents a total projection of 23 optical sections each 1.5 μm thick. The shape of individual cells was outlined using white broken lines. EM, embryonal mass; ET, embryonal tubes; ES, embryo suspensor. Scale bar corresponds to 50 μm.

(b–d) Single optical section taken through a region of the embryonal mass (b), embryonal tube(c) and a total projection of 20 individual sections, each 1.5 μm thick, taken through the embryo suspensor (d). The scale bar corresponds to 10 μm.

To further establish the role of F-actin in the development of the suspensor and the regulation of PCD, we have treated cell line 88.17 with the actin-depolymerising drugs latrunculin B (100 nm) and cytochalasin D (5 μm). The drugs were applied at the beginning of embryo formation (sampling point 3). The culture was sampled at the end of early embryogeny (sampling point 4) and analysed immediately. The latter included: (i) F-actin visualisation to confirm the effect of the drugs on the actin network, (ii) determination of the PEM to embryo composition, (iii) double staining with acetocarmine and Evan's blue to assess embryo morphology and cell viability in the embryonal mass and the suspensor and (iv) TUNEL assay in the whole culture (PEMs and embryos) to determine the frequency of nuclei with fragmented DNA.

Application of both drugs caused the de-polymerisation of F-actin in all cells (shown for latrunculin B in Figure 4a). Occasionally, F-actin fragments could be seen in some cell aggregates (data not shown). De-polymerisation of F-actin disrupted embryo pattern formation and stimulated the accumulation of PEMs (up to 99% versus 25% in the control or DMSO-treated cultures; Figure 4b).

Figure 4.

Effect of actin de-stabilisation on embryo development and PCD in cell line 88.17. EM, embryonal mass; ET, embryonal tubes; ES, embryo suspensor. Scale bar corresponds to 50 μm.

(a) An embryo after treatment with 100 nm latrunculin B, stained with rhodamine–phalloidin. Note the absence of actin filaments in all parts of the embryo.

(b) Effect of latrunculin B and cytochalasin D on embryo formation and PCD. Black and white columns (with corresponding Y-values on the left hand side) represent the proportion of PEMs and embryos in the control sample (Control) and in the samples treated with 0.1% DMSO (DMSO), 100 nm latrunculin B (LTB) and 5 μm cytochalasin D (CHD). Striped columns (with Y-values on the right hand side) represent frequency of TUNEL positive nuclei in the corresponding samples. Bars show standard errors of the mean.

(c) Images of a control embryo treated with 0.1% DMSO and an embryo treated with 100 nm of latrunculin B (LTB). Note the absence of embryonal tube cells and malformed suspensor after latrunculin B treatment. Embryos were double-stained with acetocarmine and Evan's blue.

(d) DAPI and TUNEL staining of an embryo treated with 100 nm latrunculin B. Note the appearance of TUNEL positive cells in the embryonal mass (compare with Figure 2c).

A typical image of an aberrant embryo formed after latrunculin B treatment is shown in Figure 4(c). The embryo is almost two times larger in size, has disproportionately expanded embryonal mass (compare with DMSO-treated sample) and does not form a well-defined layer of embryonal tube cells. The embryo suspensor contains a large number of small cells showing that cell elongation is inhibited (Figure 4c). In the control embryos, the suspensor cells are permeable to Evan's blue dye, whereas the cells within the embryonal mass stain red with acetocarmine (Figure 4c). After treatment with anti-actin drugs, many Evan's blue positive cells appear throughout the embryonal mass together with acetocarmine positive cells (Figure 4c), indicating that this embryonic structure contains dying cells with disrupted plasma membranes. At the same time, in latrunculin B- and cytochalasin D-treated samples, the number of cells with fragmented DNA increased by 30 and 100%, respectively (Figure 4b). Moreover, TUNEL positive cells were found not only in the suspensor region, but also throughout the embryonal mass (Figure 4d, compare with Figure 2c). Thus, de-polymerisation of actin disrupts correct embryo pattern formation through the inhibition of suspensor differentiation, which leads to embryo abortion.

Actin filaments, but not microtubules are known to be essential for both apoptotic and autophagic cell death in animals. In the case of apoptosis, microtubules degrade earlier than F-actin and the latter forms thick bundles (Coleman and Olson, 2002; Jochova et al., 1997). F-actin is required for cell contraction and membrane blebbing during the initial phase of apoptosis; it disappears later during the formation of apoptotic bodies (Coleman and Olson, 2002). In the case of autophagy, it has been shown that anti-actin drugs cytochalasin B and D inhibit the formation of autophagosomes (Aplin et al., 1992). Also, death-associated protein kinase (DAPk), which is involved in the activation of autophagosome formation during autophagic PCD, is associated with actin filaments (Cohen et al., 1997). It has been shown that this interaction is important for DAPk to influence cellular disassembly (Inbal et al., 2002).

Taken together, these data suggest that, as in autophagy and apoptosis in animal cells, the actin cytoskeleton is also an important component of the plant PCD pathway. The actin network may regulate PCD by providing the support for the components that control autophagosome formation and also direct autophagosome movements in the cytoplasm (Klionsky and Emr, 2000). Actin might also play a role as a negative regulator of autophagic PCD in the developing embryo, and the preservation of F-actin cables in suspensor cells might be a part of the regulatory machinery that controls the timing of PCD.

Role of MAP-65 in embryonic pattern formation and PCD

Microtubule-associated proteins are essential for animal embryo development. In mice, a knockout in MAP 1B results in aberrant embryogenesis and finally in embryo abortion (Edelmann et al., 1996). The plant MAP with molecular weight 65 kDa (MAP-65) is important for plant microtubule organisation in vivo and in vitro (Jiang and Sonobe, 1993; Rutten et al., 1997; Smertenko et al., 2000). We therefore compared the expression and localisation of MAP-65 in normal (88.17) versus developmentally arrested and PCD-deficient (88.1) embryogenic cell lines of P. abies (Figure 1a).

Cell line 88.1 had been selected through the screening of P. abies embryogenic cell lines induced from individual zygotic embryos based on its inability to pass through PEM-to-embryo transition, regardless of treatment (Figure 1a,c). This phenotypic feature was confirmed by time-lapse tracking analysis of the development of individual PEMs (Filonova et al., 2000b). Developmental arrest in cell line 88.1 appears to correlate with a 7–20-fold lower level of TUNEL compared to line 88.17 at the corresponding sampling time points (Figure 1d).

Microtubules in the PEMs of both cell lines were stained using anti-α-tubulin. Microtubules in the cell line 88.1 are in criss-cross orientation, whereas microtubules in cell line 88.17 are radial (Figure 5a). Staining with an antiserum raised against MAP-65 revealed further differences. In the developmentally arrested cell line, MAP-65 was not observed to be bound to cortical microtubules but instead was detected as free cytoplasmic protein as evidenced by the cytoplasmic staining (Figure 5a). This feature was observed at all five sampling time points. In the PEMs and the embryos of the normal cell line 88.17, a subset of microtubules was always decorated with MAP-65 at all sampling time points (Figure 5a). This staining pattern is similar to the MAP-65 staining of tobacco cells previously described (Smertenko et al., 2000). These data raise the possibility that the binding of the cortical microtubule cytoskeleton by MAP-65 is important for embryo differentiation. In animals, MAP1b-stabilised microtubules are essential for embryo development, as animals deficient in this MAP do not produce embryos and, moreover, those with decreased amounts of MAP1b show severe embryo abnormalities (Edelmann et al., 1996). Interestingly, MAP-65 does decorate the phragmoplast midzone in both cell lines (as shown for tobacco in Smertenko et al., 2000), indicating that MAP-65 is capable of binding microtubules in this array (Figure 6). Therefore, the putative mutation in cell line 88.1 does not affect MAP-65 binding to microtubules per se, but it prevents specific binding of MAP-65 to cortical microtubules.

Figure 5.

Localisation and expression of MAP-65 in the cell lines 88.1 and 88.17.

(a) Double staining of microtubules and MAP-65 in PEM cells in lines 88.1 and 88.17 collected at time point 1. Scale bar corresponds to 10 μm.

(b) Double staining of microtubules and MAP-65 in an early embryo in cell line 88.17 collected at time point 4. Note the weak staining of α-tubulin and no staining with anti-MAP65 in the suspensor cells. Inset is a schematic drawing of the composite cells. EM, embryonal mass; ET, embryonal tubes; ES, embryo suspensor. Scale bar corresponds to 20 μm.

(c) Western blot of a 1D gel of total protein extracts from cell lines 88.1 and 88.17 collected at five sampling time points probed with anti-NtMAP65-1 followed by anti-tubulin.

(d) Variations in the amount of 63- and 65-kDa isoforms of MAP-65 in cell line 88.1 and 88.17 at five sampling time points. Similar Western blots to the one shown in (c) were probed with four different anti-MAP-65 antibodies and the amounts of the 63- and the 65-kDa isoforms were quantified using densitometric scanning. The graphs show the combined data obtained with all four antibodies. Bars show standard errors of the mean.

Figure 6.

Localisation of MAP-65 during cytokinesis in PEM cells in cell lines 88.1 and 88.17 collected at time point 1.

(a–c): Cell line 88.1; (d–f): cell line 88.17 double stained for α-tubulin (a,d), and MAP-65 (b,e). (c,f): merged image; yellow colouration indicates co-localisation. In both cell lines MAP-65 localises to the phragmoplast midzone. Scale bar corresponds to 10 μm.

In P. abies somatic embryos from line 88.17, MAP-65 was bound to microtubules in all the cells within the embryonal mass and to the transverse microtubules in the embryonal tube cells (Figure 5b). However, in some embryonal tube cells and in all suspensor cells the microtubule network was disrupted with only the fragments of microtubules remaining in the cytoplasm (Figure 5b). As MAP-65 was present in the cytoplasm of these cells but was not bound to the remnants of the microtubules (Figure 5b), it becomes evident that MAP-65 is only bound to organised microtubules as previously suggested (Smertenko et al., 2000). It should be noted that line 88.1 does not form somatic embryos, and at the same sampling time point as the somatic embryo from line 88.17 (Figure 5b) cells from line 88.1 are very similar to cells in Figure 5(a).

Antibodies raised against recombinant tobacco NtMAP65-1a identify two electrophoretically separable isoforms on 1D gel immunoblots (the lower band of which is often a doublet; Smertenko et al., 2000). This antibody cross-reacts with similar bands of 63 and 65 kDa on 1D gel immunoblots of total protein extracts from the two P. abies cell lines (Figure 5c). We have compared the relative abundance of the two MAP-65 isoforms in both cell lines at the five successive sampling time points (Figure 5c). Similar Western blots to those shown in the Figure 5(c) were also probed with antibodies against the biochemically purified tobacco MAP-65 family of proteins (Jiang and Sonobe, 1993) and with two antisera against recombinant Arabidopsis thaliana MAP-65 proteins (A.S. and P.J.H., unpublished data). The relative amounts of 65- and 63-kDa isoforms at each sampling time point on all four Western blots were quantified using densitometric scanning and the results are plotted in Figure 5(d).

In cell line 88.1, the amount of both MAP-65 isoforms is invariant at all sampling time points in that the 63-kDa isoform was constantly lower than the 65-kDa isoform. In contrast, in cell line 88.17 the relative levels of the two MAP-65 isoforms progressively changed as the ratio of somatic embryos to PEMs increased in the cell culture (Figures 1c and 5c,d). The quantity of the 63-kDa isoform varies as embryos accumulate but the 65-kDa protein steadily decreases. At time point 5, where the sample has only embryos, the 65-kDa isoform is barely detectable (Figure 5c,d). In addition, anti-α-tubulin was used on the immunoblot as a control and no statistically significant variation in the signal in either line at any of the sampling time points was observed (Figure 5c).

Taking the immunofluorescence and isoform analysis data for the two cell lines together, we suggest that one of the criteria for embryonic pattern formation is the presence of cortical microtubules decorated by MAP-65. The observation that organised microtubules are bound by MAP-65 at all sampling time points in cell line 88.17 would imply that a critical quantity of 63-kDa isoform is essential for cell differentiation. This might indicate a similar necessity for the 63-kDa MAP in plant embryogenesis as MAP1b is essential for animal embryo development (Edelmann et al., 1996).

Model for cytoskeletal re-organisation during embryonic PCD in Picea abies

The process of PCD can be divided into two major phases. During the first phase, commitment phase, the cell makes the decision on whether to enter the PCD pathway. This decision depends on the pattern of gene expression and extracellular signals. The second phase is the execution of PCD, when the cell contents are sequentially destroyed. The signal that makes cells commit to PCD in the P. abies embryo and how this signal is transduced is not known.

Ultrastructural analysis of the different cell types in P. abies somatic embryos has previously shown no signs of cell degradation in the embryonal mass (Filonova et al., 2000a). The first signs of PCD execution were observed in the embryonal tube cells: the number of Golgi increased and autolytic provacuoles appeared (Filonova et al., 2000a). These cells originate through asymmetric division of the basal cells within the embryonal mass and have very low division activity, as we could not find any dividing cells among embryonal tube cells in 75 individual embryos analysed in this work. Two types of microtubule organisation were observed in this intervening part of the embryo: transverse with MAP-65 bound to microtubules and disorganised with MAP-65 not bound to the microtubule fragments (Figures 2d and 5b). As none of the embryonal tube cells with a partially disorganised microtubule network contained bound MAP-65, we conclude that the dissociation of MAP-65 from microtubules occurs concomitantly with the disorganisation of microtubules. It is possible that the beginning of the PCD execution pathway coincides with the re-orientation and the subsequent disorganisation of microtubules. F-actin was found in the embryonal tube cells (Figure 3c), so it is clear that this component of the cytoskeleton remains intact at the onset of the execution phase of PCD.

The subsequent events in the PCD execution pathway involve the transformation of autolytic provacuoles into autolytic vacuoles that grow in size and destroy the cytoplasmic content (Filonova et al., 2000a). The cells at this stage of PCD can be found both in the embryonal tube and the suspensor regions. Suspensor cells are largely elongate and can be up to 20-fold larger than embryonal tube cells. It is tempting to speculate that the suspensor cells are unable to maintain an isodiametric shape owing to the collapse of the microtubule network, and that the direction of the cell expansion is regulated by F-actin which is re-arranged in oblique and longitudinal cables (Figure 3d). Application of actin-depolymerising drugs suggests that F-actin controls both suspensor patterning and PCD (Figure 4).

Further during the execution phase of PCD, the DNA fragments (Figure 2c,e), the large central vacuole collapses and the whole content of the cell is removed, leaving only a hollow walled-cell corpse (Filonova et al., 2000a). The cells at this culminate stage of PCD are located at the distal end of the embryo suspensor (Figure 2c).

In order to unravel the biochemical pathways of plant PCD, it is of utmost importance to understand the sequence of the cytomorphological changes underlying cell disassembly. In Figure 7 we have drawn a scheme for cytoskeletal re-organisation during embryonic PCD in P. abies. This scheme is based on the combined data from the previous studies on embryo pattern formation (Filonova et al., 2000b) and associated PCD (Filonova et al., 2000a) together with those presented here.

Figure 7.

Schematic model for the re-organisation of the cytoskeleton in the course of PCD associated with embryonic pattern formation in Picea abies.

Different colours highlight three major cell types that build up the apical–basal pattern of the embryo: green, embryonal mass cells; yellow, tube cells; orange, suspensor cells.

Stages of PCD: 0, proliferating cells, no features of PCD; I, commitment to PCD; II–V, execution of PCD; II, formation of autolytic vacuoles; III, growth of vacuoles; IV, DNA fragmentation; V, cell corpse.

The scheme includes six major stages of cell disassembly. Stage 0 applies to living meristematic cells of the embryonal mass which will differentiate into a mature embryo through strictly co-ordinated cell divisions (Filonova et al., 2000b). The cells within the embryonal mass contain dense cytoplasm, a small number of vacuoles and a rounded nucleus which occupies a substantial proportion of the protoplast (Filonova et al., 2000a). F-actin and microtubules in these cells form fine networks (Figures 2d and 3b); MAP-65 is bound to a subset of microtubules (Figure 5b).

Stage I involves the embryonal tube cells that show an increased number and activity of Golgi complexes and an accumulation of small autolytic provacuoles in the cytoplasm (Filonova et al., 2000a). These cells are usually elongate with the microtubules oriented transverse (Figure 2d), and again MAP-65 is bound to a subset of these microtubules (Figure 5b). Cells at stage I are in the commitment phase of PCD.

The execution phase of PCD includes stages II to V with the cells located in the embryonal tube and in the suspensor zones. During stage II, MAP-65 dissociates from the microtubules (Figure 5b) and the microtubule network is disrupted (Figures 2d and 5b). The cells at stage II are much larger than the cells at the previous stages and they expand further as they progress towards stage III. During stage II, the vacuoles expand either as a result of autophagocytosis where one vacuole acquires another or by vacuole fusion (Filonova et al., 2000a). By stage III, several large vacuoles occupy the majority of the cell volume. No microtubules are left by this stage and only microtubule fragments can be seen in the cytoplasm (Figures 2d and 5b). In contrast to microtubules, F-actin forms thick longitudinal bundles at both stages II and III (Figure 3d). At stage IV, the nuclei are TUNEL positive. Once the vacuole collapses, the remaining cytoplasm is degraded leaving a hollow walled-cell corpse (stage V; Filonova et al., 2000a).

In conclusion, our data indicate that microtubules and F-actin are both important for embryogenesis, but it is F-actin that is a critical for the execution of embryonic PCD. Furthermore, the results presented here have a wider significance beyond just plant embryology as most developmental cell deaths in plants recruit a pathway of cell dismantling similar to the one defined in P. abies embryos (Beers and McDowell, 2001; Filonova et al., 2000a, 2002; Zhivotovsky, 2002). Further work will involve trying to understand the nature of the PCD signal in the embryo and how this is transduced to the cytoskeleton.

Experimental procedures

Cell cultures

Two embryogenic cell lines of P. abies (Norway spruce), 88.1 and 88.17, were used in this study. Both cell lines were stored in liquid nitrogen and thawed 5 months prior to the onset of experiments. The treatments for PEM proliferation (+PGR), PEM-to-embryo transition (−PGR) and embryo maturation (+ABA) were conducted as previously described (Bozhkov et al., 2002; Filonova et al., 2000a). The samples for Western blotting as well as individual PEMs and/or somatic embryos for immunofluorescence microscopy were collected sequentially at five critical time points: 3 and 7 days after addition of PGRs, 24 h and 7 days after withdrawal of PGRs and 7 days after the start of ABA treatment. The percentage composition of PEMs and somatic embryos in both cell lines at all sampling time points was determined after observation of at least 500 individual cell aggregates with at least four replicates on every occasion (Filonova et al., 2000a).

In situ detection of DNA fragmentation (TUNEL assay)

Terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick-end labelling and DAPI counterstaining of whole mount PEMs (cell lines 88.1 and 88.17) and early somatic embryos (cell line 88.17) were performed as previously described (Filonova et al., 2000a), the only difference being that a TMR red in situ cell death detection kit (Boehringer Mannheim) was used in this study. To assess the frequency of nuclei with fragmented DNA, at least 1000 nuclei were observed in TUNEL-stained PEMs and/or embryos at distinct time points, with at least three replicates on every occasion. Application of TMR red labelled-dUTP without TdT was used as the negative control. For double visualisation of TUNEL positive nuclei and tubulin, the samples were prepared in the following order: (i) fixation with paraformaldehyde, (ii) incubation with TUNEL reaction mixture containing TMR red labelled-dUTP and TdT, (iii) treatment with the cell wall digesting enzymes and (iv) incubation with mouse anti-α-tubulin antibody followed by secondary antimouse FITC-conjugated antibody.

SDS–PAGE and Western blotting

Proteins were extracted from fresh material of cell lines 88.1 and 88.17 collected at the five sampling time points. In the preliminary experiment, two types of protein extracts were tested: total protein and soluble protein. Both extracts gave identical results and for all experiments we used soluble protein extracts. After separation on 7.5% pre-cast gels (Bio-Rad), proteins were transferred onto nitrocellulose membrane and probed with antibodies (anti-α-tubulin and MAP65 antibodies) as described in Smertenko et al. (1997).

After development, the membranes were scanned and the bands quantified with the Quantity One Image Analysis Program (Bio-Rad Ltd). Local values for background were estimated using the program after initial testing with global values. For each replicate (membrane hybridised with a separate MAP antibody), the pixel values (intensities) for each band were adjusted so that the lowest scoring intensity was zero. Then the data from the various replicates were normalised by mean: the intensity for each band in a particular replicate was divided by the mean intensity for all bands in the replicate so that values were expressed on a scale where value ‘1’ was the mean intensity. The data were then averaged across four replicates.

Immunofluorescence microscopy

The staining of PEMs and embryos was performed using the method in Smertenko et al. (1997). Briefly, cell aggregates were fixed for 40 min in freshly prepared 3.7% paraformaldehyde (Sigma, Dorset, UK; Cat. No. P-6148) solution in PEM buffer (50 mm PIPES, 2 mm EGTA and 2 mm MgSO4). After fixation, the cells were washed twice in PBS, attached on to the poly-l-Lysine coated coverslips and subjected to the cell wall digestion for 10 min with 1% Macerozyme R-10 and 0.2% Pectolyase Y-23 diluted in 0.4 m mannitol, 15 mm MES, pH 5.0, 5 mm EGTA, 1 mm PEMS and 0.01 mg ml−1 of leupeptin. After that, the cells were stained with rabbit polyclonal anti-MAP65 antibodies, monoclonal anti-α-tubulin (N356, Amersham, High Wycombe, UK) and anti-actin (clone C4, ICN). As secondary antibodies, anti-rabbit FITC (Sigma, Poole, UK) conjugates and anti-mouse TexasRed conjugates (Amersham, High Wycombe, UK) were used.

The F-actin was also visualised with rhodamine–phalloidin (Molecular Probes) using the method described by Binarova et al. (1996). The samples were incubated in 1 μg ml−1 solution of DAPI in PBS to visualise DNA, mounted in the Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA) and examined using a Bio-Rad Radian 2000 confocal microscope.

Both methods of F-actin visualisation (using anti-actin or rhodamine–phalloidin) detected the same staining pattern throughout the embryos.

Drug application

To assess the effects of actin de-stabilisation on embryogenesis and PCD, cell line 88.17 was treated with 100 nm latrunculin B (Calbiochem) and 5 μm cytochalasin D (Sigma). Both drugs were dissolved in DMSO and added to the suspension culture 24 h after withdrawal of PGRs (time point 3). The final concentration of DMSO in the medium was 0.1% (v/v). Untreated cultures and those treated with 0.1% (v/v) DMSO were used as controls.

The control and drug-treated cell samples were collected at 7 days after withdrawal of PGRs (time point 4) and processed for immediate analyses. The latter included: (i) determination of the percentage composition of PEMs and embryos (as described above), (ii) double staining with acetocarmine and Evan's blue to assess embryo morphology and cell viability in the embryonal mass and the suspensor as described in Filonova et al. (2000b), (iii) TUNEL assay to estimate the frequency of nuclei with fragmented DNA (as described above), and (iv) rhodamine–phalloidin staining of F-actin (as described above). At least 100 embryos per treatment were analysed after staining with both acetocarmine/Evan's blue and rhodamine–phalloidin. All the analyses were repeated three times.


We wish to thank Seiji Sonobe for the tobacco anti-MAP-65 antiserum. This work was supported by The Royal Swedish Academy of Agriculture and Forestry, The Troedssons Fund (Sweden), and the Biotechnology and Biological Sciences Research Council (UK).