Immunological characterization of lamins in the nuclear matrix of onion cells

The complex family of intermediate filament (IF) proteins in vertebrates comprises more than 40 different proteins of variable molecular mass and specific expression, classified into five families. Even though they are functionally very different, all of them share a very conserved molecular structure with a central α-helix rod domain with typical coiled-coil segments, however the N-terminal heads and C- terminal tails are very variable (Osborn and Weber, 1986; Franke, 1987) and form homodimeric complexes. All the IFs tested contain a common epitope at the C-terminal end of the helical rod domain, which is recognized by a monoclonal antibody named IFA (Pruss et al., 1981). It is now known that IF proteins are not exclusive to vertebrates, and they have been characterized in several invertebrates (Dodemont et al., 1990; Riemer et al., 1991).

Although IFs have not been described as such in plants, the IFA antigen appears to be widely distributed in them (Dawson et al., 1985; Goodbody et al., 1989; Frederick, et al., 1992; Li and Roux, 1992), and IF epitopes have been reported in the cytoskeletons of several plant species (see Shaw et al., 1991). In addition, crossreactivity has been demonstrated between plant and animal IFs. The data available suggest that the epitopes detected on plants are generic and conserved phylogenetically (Shaw et al., 1991). The information on plant IFs is scarce. Plant cells do not have an organized cytoskeletal network of IFs, although the presence of IF antigens in both fibrillar bundles (FBs), which are capable of forming 10 nm filaments after denaturation, and microtubule-associated arrays have been demonstrated (Goodbody et al., 1989; Hargreaves et al., 1989b).

The nuclear lamina is a scaffolding structure that corresponds to the residual elements of the nuclear envelope, which has been, so far, described in eukaryotes ranging from Protozoa to vertebrates, as being a universal component of eukaryotic nuclei (Gerace, 1986; Krohne and Benavente, 1986; Doring and Stick, 1990). The lamina is made up basically of a network of lamin filaments with which the residual elements of the pore-complexes are associated (Krohne and Benavente, 1986; Aebi et al., 1984), and shows continuity with the cytoskeleton of intermediate filaments (Capco et al., 1984; Carmo-Fonseca et al., 1987). The major structural proteins of the lamina are the lamins, a multigenic family of proteins that share many structural features with the IF proteins and are classified as type V IF (McKeon et al., 1986; Osborn and Weber, 1986; Franke, 1987). Lamins appear to be ubiquitous in eukaryotic nuclei, and they have been reported in yeast (Georgatos et al., 1989), invertebrates (Dessev et al., 1990) and vertebrates (Krohne and Benavente, 1986). IF of invertebrates apparently show a closer relationship to lamins than their vertebrate counterparts (Weber et al., 1989; Dodemont et al., 1990). Nevertheless nuclear lamins have some characteristics that distinguish them from the IF proteins, such as: the presence of a nuclear localization signal and the C-terminal sequence motif CaaX (Loewinger and Mckeon, 1988).

According to their antigenic determinants, sequence analysis, specific expression and subcellular distribution in mitosis, several lamin subtypes have been characterized: lamins A, B and C in mammals, lamin A, B₁ and B₂ in chicken, and up to five different lamins (L₁ to L_IV and LA) in Xenopus (Stick, 1988). All of them have been classified in three different groups: type-A, type-B and L_III (Stick, 1988).

The existence of a lamina in the nuclear matrix of onion cells has been well documented by electron microscopy (EM) (Cerezuela and Moreno Díaz de la Espina, 1990; Moreno Díaz de la Espina et al., 1991). But up to now the presence of lamins in this lamina had not yet been proved conclusively, although crossreaction between some plant nuclear matrix proteins and anti-IF antibodies from different sources has been reported (Galcheva-Gargova et al., 1988; Moreno Díaz de la Espina et al., 1990; Beven et al., 1991; Mínguez and Moreno Díaz de la Espina, 1992; Frederick et al., 1992; Li and Roux, 1992; McNulty and Saunders, 1992). Since higher plants have cytoplasmic IF antigens that are not organized into a cytoskeletal framework (Dawson et al., 1985; Hargreaves et al., 1989a; Shaw et al., 1991), the demonstration of lamins organizing a lamina on them, as well as the determination of their subtype, would be of great interest in the clarification of the evolution of the IF proteins in eukaryotes.

We report here the presence of homologues of vertebrate lamins in the lamina of a higher plant, the onion. Our results, in addition to those confirming the existence of lamins in lower eukaryotes (Georgatos et al., 1989; Mínguez et al., unpublished data) and other higher plants (Galcheva-Gargova et al., 1988; McNulty and Saunders, 1992) as well as the close relationships in the molecular structure found between lamins and IF of invertebrates, support the hypothesis of a lamin-like ancestor for lamins and IF proteins (Doring and Stick, 1990; Dodemont et al., 1990; Beven et al., 1991).

The materials used were root meristem cells from Allium cepa L. bulbs grown in filtered tap water under constant temperature conditions (15(± 0.5)°C).

This was performed as previously described (Moreno Díaz de la Espina et al., 1991).

Nuclear matrices were prepared by five successive extractions of nuclei in 10 mM Tris-HCl, pH 7.4, containing in each case: (1st) 50 μg/ml DNase I and 5 mM MgCl₂; (2nd) 0.5% Triton X-100, 5 mM MgCl₂; (3rd) 200 μg/ml DNase I, 200 μg/ml RNase A, 5 mM MgCl₂; (4th) 0.25 mM MgCl₂ and (5th) 2 M NaCl, 0.25 mM MgCl₂. In some experiments, nuclear matrices were further extracted in 4 M urea in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA. Immediately before use 1 mM PMSF and 20 mM DTT were added to all extraction media. Sedimentation of nuclei and monitoring for DNA elimination from nuclear structures were performed as previously described (Moreno Díaz de la Espina et al., 1991).

One-dimensional (1-D) SDS-PAGE was performed in 10% acrylamide gels according to Laemmli (1970), as previously described (Moreno Díaz de la Espina et al., 1991).

Two-dimensional (2-D) SDS-PAGE was performed according to O’Farrell (1975), using pharmalytes, pH 3-10. Samples were solubilized at 100°C for 5 min in a sample solution containing 8 M urea, 2% 2-mercaptoethanol, 2% pharmalytes 3-10 and 0.5% Triton X-100. The second dimension was in 10% acrylamide gels. Gels were stained with silver.

To avoid crossreactivity with cytoplasmic IF-type proteins either nuclei or nuclear matrix fractions were used instead of whole cell proteins for immunoblotting. After electrophoresis samples were transferred to nitrocellulose membranes using a semi-dry electric blotter (Millipore).

The following antibodies and dilutions were used: a rabbit polyclonal against chicken lamins (1:75) (Stick and Hausen, 1980). L32B8 a mouse monoclonal against chicken lamin B₂ (undiluted) (Stick et al., 1988). Three monoclonals against Xenopus lamins: L7-4A2 (L_I) (1/100), L78C6 (L_II) (1/100) and L65DC (L_II and L_III) (1/100) (Stick, 1988). The IFA recognizing a very conserved epitope in all types of IF proteins (undiluted) (Pruss et al., 1981) anti-vimentin (undiluted) (Oncogen Sciences), AFB (undiluted) (Hargreaves et al., 1989a) and MAC 322 a monoclonal against carrot cytoskeletons recognizing a cytokeratin 8 epitope shared by higher plants and animals (Ross et al., 1991), an anti-Sm recognizing an epitope common to plant and animal snRNP particles (Vázquez Nin et al., 1992). The anti-lamin serum was purified by immunoaffinity on nitrocellulose-bound antigen according to Krohne et al. (1982).

After being washed in 0.05% Tween-20 in PBS and blocked with a 3% solution of dried non-fat milk in distilled water, blots were incubated with the primary antibodies at different dilutions. Second antibodies, labelled with peroxidase and diluted at 1:500 (Sigma) were used; the reaction was revealed by chloronaphthol. The reaction with the monoclonal anti-lamin antibodies was revealed by the enhanced chemiluminiscence ECL western blotting system from Amersham. In this case the second antibodies, diluted at 1/10000, were used.

As a positive control, proteins from a culture of Escherichia coli that expresses Xenopus L_III were blotted against the anti-lamin serum. Negative controls were performed by omitting the primary antibodies or substituting them with unrelated antibodies recognizing nuclear antigens.

The rabbit polyclonal antibody against chicken lamins (Stick and Hausen, 1980) was used as a primary antibody. A second FITC- conjugated sheep anti-rabbit antibody was used. To avoid interference with the autofluorescence of plant cells and also to favour penetration of the antibodies without cell walls, isolated nuclei and nuclear matrices were used for immunofluorescence experiments.

Nuclear and nuclear matrix fractions were fixed in 0.3% PFA in PBS, pH 7.0, washed in the same buffer, spread on polylysine- coated slides and air dried. After drying, slides were quickly dipped in methanol-acetone (1:3, v/v) at −20°C.

After blocking with preimmune sheep serum, incubation with the primary antibody at 1:50 dilution in PBS was carried out for 30 min at 37°C. After washing in PBS, slides were incubated with the second antibody at 1:400 dilution. After washing, preparations were mounted with an antifading medium containing 90% glycerol, 0.01% paraphenylendiamine in PBS and 15% Hoechst to counterstain DNA. Negative controls were performed by incubating with normal rabbit serum instead of the primary antibody. The reaction with the antibody was observed under an epifluorescence microscope using a 450-490 nm filter and the DNA-fluorochrome complex at 365 nm.

Pellets from nuclei and nuclear matrices were fixed in 4% PFA in PBS, pH 7.2, for 2 h at 4°C and embedded without postfixation in LR white acrylic resin as previously described (Martin et al., 1989).

Postembedding immunolabelling was performed on ultrathin sections mounted on nickel grids as previously described (Martin et al., 1989). The anti-lamin antibody was used at different dilutions from 1:50 to 1:1000. A gold-conjugated goat anti-rabbit (Janssen) was used as second antibody at 1:500 dilution. Control experiments were done by omitting the incubation with the first antibody or incubating with purified rabbit IgG.

The distribution of labelling was quantified by using 30 micrographs from randomly chosen matrices taken from three different incubations and from two different grids from each experiment. The final magnification was ×65,000. Particle densities in the three main matrix components were calculated using a semiautomatic procedure in a Tandom Target 585 6 × 116 MH₂ computer with a digitizer tablet.

The plant nuclear matrix has a complex polypeptide profile in 1-D SDS-PAGE gels (Fig. 1) (Moreno Díaz de la Espina et al., 1991). Our efforts to eliminate the internal components of these matrices and produce the pore complexlamina fraction by using different experimental procedures were unsuccessful (Cerezuela and Moreno Díaz de la Espina, 1990). Extraction with 4 M urea did not completely remove the internal matrix, as proved by the electron and light microscopic images, but eliminated many of its protein components (Fig. 2A,B), hence we used this treatment to relatively concentrate the putative lamin bands in the nuclear matrix fraction.

Two-dimensional gels of nuclear matrix fractions show many spots between 70 and 40 kDa with a variable range of pI values. Low molecular mass proteins are few and either acidic or neutral (Fig. 2A). Urea extraction removes a lot of proteins from the matrices, especially acidic proteins, in such a way that urea-extracted nuclear matrices present few distinct spots with mostly neutral and basic pi values (Fig. 2B).

When the anti-chicken lamin serum was titered against the nuclear matrix proteins from onions at concentrations below 1:75 dilution, there was a single band at about 65 kDa in 1-D blots that is enriched in urea-extracted nuclear matrices. This band is also recognized by the iFA antibody reacting with a very conserved epitope of the carboxylic end of the conserved domain of iF proteins (Fig. 1), which also reveals an additional band at about 58 kDa. Sometimes additional bands of iF proteins with lower molecular mass values appear, probably due to proteolytic degradation during sample preparation (Fig. 1).

The chicken anti-lamin serum shows crossreactivity with a bacterially expressed lamin L_iii of Xenopus laevis, suggesting the presence of common epitopes between avian, plant and amphibian lamins (Fig. 1).

By using the ECL developing system, we demonstrated the reaction of the 65 kDa onion lamin with monoclonal antibodies specific for chicken lamin B₂ (L32B8) and lamins L_i (L7 4A2) and L_ii (L78C6) from Xenopus, which have been classified as belonging to the B lamin subtype (Fig. 3).

To eliminate the possibility of contamination of the fractions with cytoplasmic iF type antigens, we tested different antibodies against iFs: anti-vimentin, AFB and MAC 322; the latter recognizing a very conserved epitope of iF shared by plant and animal cells. None of them recognizes the 65 kDa band corresponding to lamins (Fig. 1, and data not shown). This band was not recognized by other antibodies against abundant nuclear antigens unrelated to iF, as were the Sm proteins of snRNP particles (Vázquez-Nin et al., 1992, not shown here).

In 2-D blots of nuclear matrices the iFA antibody recognizes several spots corresponding to iF-type proteins. The most abundant has a molecular mass of 58 kDa and a PI of 5. The rest have molecular masses very close to each other at about 65 kDa with pi values between 6.8 and 6.1 (Fig. 2D).

The polyclonal anti-chicken lamin serum recognizes one of these spots at 65 kDa and pi 6.8 (Fig. 2C and D). The L74A2 mAb against Xenopus L_I reacts with a different spot at 64 kDa and pi 5.75 (Fig. 3E).

When isolated nuclei and nuclear matrices were incubated with the anti-lamin antibody and observed by indirect immunofluorescence the nuclear rims appeared to be decorated (Fig. 4A, B). The penetration of the antibodies into the nuclear interior in the conditions of fixation and per- meabilization used was tested by using an anti-topoiso- merase I human serum, which recognizes the enzyme in plants (data not shown here). Nuclear matrices, in addition to the decoration of their peripheries, show a spot-like decoration (Fig. 4B). The fluorescence is neither related to the chromatin, as proved by the fluorescence with Hoechst (Fig. 4D, E and F), nor to the nucleolar matrix, because there are fluorescent spots in areas corresponding mainly to the nucleolus as shown by phase-contrast microscopy and Hoechst staining (Fig. 4G and H). Incubation with normal rabbit serum does not produce any decoration (Fig. 3C).

After postembedding labelling with the anti-lamin serum, the gold particles mainly decorate the lamina and the internal matrix, but not the residual nucleolus (Fig. 5A). Quantification of the data shows that the specific reaction is distributed between the lamina and the internal matrix (Fig. 6). There was no labelling when the incubation with the primary antibody was omitted (Fig. 5B).

In contrast with the situation in the plant cytoskeleton, in which IF-related antigens do actually exist, but a well-organized network of IF-type filaments has never been demonstrated (see Shaw et al., 1991). Our previous studies with the electron microscope revealed the existence of a well- organized lamina with associated pore complexes in the onion nuclear matrix (Barthelemy and Moreno Díaz de la Espina, 1984; Cerezuela and Moreno Díaz de la Espina, 1990; Moreno Díaz de la Espina et al., 1991), homologous from a structural point of view to that found in other eukaryotes (see Krohne and Benavente, 1986). This observation was later extended to other plant systems such as maize (Krachmarov et al., 1991), carrot (Frederick et al., 1992), pea (Li and Roux, 1992) and beans (Galcheva-Gar- gova et al., 1988), in such a way that the nuclear lamina is now considered to be a ubiquitous structural component of the nuclei of higher plants.

Although the presence of a well-developed lamina in plant nuclear matrices points to lamins as putative components of the matrices, until now these proteins could not be definitely demonstrated in plant systems.

Immunological analysis has demonstrated that plant nuclei contain IF-related antigens (Galcheva-Gargova et al., 1988; Beven et al., 1991; Shaw et al., 1991; Frederick et al., 1992; Li and Roux, 1992) but it has not yet been proved whether these proteins do actually correspond to lamins.

When antibodies specific for vertebrate lamins have been used in plant cells the results were rather uncertain and proteins with molecular mass values as different as 55 and 92 kDa, showing an intranuclear, not only a peripheral, distribution, were described as putative plant lamins in view of their crossreactivity with different anti-lamin antibodies (Barthelemy and Moreno Díaz de la Espina, 1984; Galcheva-Gargova et al., 1988; Beven et al., 1991; McNulty and Saunders, 1992; Li and Roux, 1992).

This confusion is probably caused by two reasons. First the lack of monospecific anti-lamin antibodies showing a good reactivity with plant lamins, and second the set of nuclear IF-type proteins in plants could actually be more complex than we think, including other families of proteins distinct from lamins (see Yang et al., 1992) but which share some antigenic determinants with them and could account for the singular topological localization reported for plant lamins (McNulty and Saunders, 1992).

Our results using polyclonal and monoclonal antibodies against vertebrate lamins from different sources (Table 1) provide strong evidence for the presence of lamins in the plant nuclear matrix and minimize the fortuitous crossre- action with unrelated proteins. The reaction of all these proteins with the IFA antibody recognizing a conserved epitope that appears to be essential for the assembly of a normal IF network (Hartzfeld and Weber, 1992), and is found in all the lamins so far studied (Osborn and Weber, 1987), not only confirms these proteins as members of the poorly studied IF family of proteins in plants (Shaw et al., 1991), but also their capability to form a network.

Contamination of the NM fraction with cytoplasmic IF- type antigens is discounted not only on the basis of microscopic observations, but also because the putative lamins do not crossreact with antibodies against other IF antigens of plant cells like AFB (Hargreaves et al., 1989a) and MAC 322, which recognize a very conserved epitope of IF shared by plant and animal cytoskeletons (Ross et al., 1991).

The IFA antibody reveals five different IF-type proteins in the onion nuclear matrix, two of which show specific reactivity with different anti-lamin sera. The polyclonal serum against chicken lamins recognizes a very conserved epitope of lamins as proved by its crossreaction with lamins from species that are very distant from each other in evolution like rat liver (Stick and Hausen, 1980), Xenopus L_III, the onion lamin and a lamin in a Dinoflagellate nuclear matrix (Mínguez et al., unpublished data), suggesting that this epitope is shared by higher and lower eukaryotes. The serum is also very reactive and specific in the onion, recognizing a single spot of the five that show reactivity with IFA.

Two-dimensional blots reveal that the polyclonal serum and the monoclonal against Xenopus L_I recognize different proteins, in spite of their similar migration in 1-D gels.

The 64 kDa protein shows the most acidic isoelectric point of the onion lamins, and migrates as a single spot in non-equilibrium pH-gradient electrophoresis, in contrast to type A lamins, which usually show multiple phosphorylated forms. Due to these features and also because it is immunologically related to B-type lamins from different vertebrates we identify this protein as a B-type lamin.

The 65 kDa protein recognized by the polyclonal serum is more basic and belongs to a group of spots that probably correspond to isoforms of the same protein, both characteristic of A-type lamins. But unfortunately we could not obtain any antibody against A-type lamins that reacted in the onion to confirm the presence of this type of lamin. For this reason, although we have evidence to claim that B-type lamins, which are considered to be the ancestral lamins (Stick, 1992), are conserved components of plant nuclear matrices, additional sequencing experiments will be necessary to clarify the molecular structure of plant lamins and classify them accurately as components of the already defined lamin subtypes (Stick, 1992).

The identification of lamins in plants, in which cytoplasmic intermediate filament antigens are not organized into a cytoskeletal framework (Shaw et al., 1991), is very interesting, in view of the tight structural and phylogenetic relationships discovered between the two families of proteins (Doring and Stick 1990; McKeon et al., 1986; Franke, 1987; Weber et al., 1989).

The molecular masses of the onion lamins are in the range of those reported for other IF-type proteins detected in the plant nuclear matrix by crossreaction with the IFA antibody in carrot (Frederick et al., 1992) and pea (Li and Roux, 1992), and with anti-lamin B antibodies (McNulty and Saunders, 1992), while the bands reported to crossreact with anti-lamin A/C type antibodies described in plants show a disparity in molecular mass (Galcheva-Gargova et al., 1988; Beven et al., 1991).

At the moment we cannot decide whether the other IF- type proteins of the nuclear matrix do actually correspond to lamins. The experiments in progress, with antibodies recognizing the two different subtypes of vertebrate lamins, will probably clear up this point.

Although the immunofluorescence pattern of the plant lamin in nuclear matrices is very similar to that shown by lamins in vertebrates (Krohne and Benavente, 1986), with a spot-like peripheral localization that could reflect the discontinuous distribution of lamins in the nuclear periphery (Paddy et al., 1992), immunogold labelling on sections also reveals an intranuclear distribution of lamins.

The specificity of the labelling in 2-D blots and the quantitative results of the immunogold labelling on sections suggest a distribution of the lamin (or some antigenically related proteins) throughout the nuclear matrix except in the residual nucleolus, in contrast with the typical peripheral distribution so far described for vertebrate lamins (Krohne and Benavente, 1986), but in agreement with the scarce results of detection of IF-type antigens in plant nuclear matrices in situ (Frederick et al., 1992; McNulty and Saunders, 1992). This pattern could suggest that the set of nuclear IF-type proteins in plants could actually be more complex than we think, including other families of related proteins that share some antigenic determinants with lamins (see Yang et al., 1992), or that there could be a different organization of the nucleoskeleton in plants, in the same way that occurs with the cytoplasmic IF proteins and the cytoskeleton (Goodbody et al., 1989). But also our results could reflect the association of lamins with the intranuclear matrix, in accordance with the experiments of microinjection of lamin A in cultured cells (Goldman et al., 1992; Lutz et al., 1992), and with the distribution of lamin antibodies in synchronized cultured cells (Bridger et al., 1993).

In conclusion, we demonstrate two lamins in the lamina of onion cells with conserved values of molecular mass and pI, which appear to be distributed not only in the lamina but also in the internal elements of the matrix. Although further work is necessary for the molecular characterization of plant lamins and their organization in the lamina, our results suggest that, except for the topological distribution of lamins, the organization of the lamina and nuclear matrix is very similar in higher eucaryotes, which supports the idea that lamins are very old members of the IF proteins and have been conserved in plants better than their cytoplasmic counterparts (Doring and Stick, 1990). This opens the door to the study of the organization and peculiarities of the plant nucleoskeleton and its relationships with the cytoskeleton.

The authors are indebted to Dr R. Stick for helpful suggestions and the generous gift of the polyclonal and monoclonal anti-lamin antibodies, and Dr D. Fairbairn for the AFB, MAC322 and IFA antibodies. We thank Mrs M. Carnota and Mrs N. Fonturbel for technical assistance, Mrs V. Lafita for typing the manuscript and Mrs B. Ligus Walker for revising the English style. This work has been supported by the CICYT (project PB91-0124). A. Minguez is a recipient of a fellowship from the PFPI (Spain).