Anionic groups on the outer surfaces of isolated rat liver nuclei were rendered visible in the electron microscope by staining with colloidal iron hydroxide at different pH values. At pH 1·8 the nuclei did not adsorb particles of stain, although plasma membranes left in the same preparation showed heavy labelling. After pretreatment with neuraminidase at pH 6 the plasma membranes were no longer stained. At pH 3·0 the nuclear surfaces also stained intensely. The staining pattern acquired at this pH did not appear to be changed by neuraminidase pretreatment.
With the staining method used, rat liver nuclear surfaces seemed to have no exposed sialic acid under isolation conditions which preserve the nuclear membranes and leave the ribosomes attached to the nuclear surface. However, at higher pH values other anionic groups seem to become dissociated and are stained with colloidal iron hydroxide.
Cells have been shown to carry various anionic groups on their surface membranes, as demonstrated by biochemical (Winzler, 1970), cell-electrophoretical (Doljanski & Eisenberg, 1965; Weiss, 1969), and ultrastructural methods (Benedetti & Emmelot, 1967; Weiss & Zeigel, 1972). It has been suggested that sialic acid, ribonucleic acid (Weiss, 1969) and in some cells sulphated groups (Marx, Graf & Wesemann, 1973) contribute to the negative surface charge observed on the surfaces of these cells. The presence on plasma membranes of macromolecules containing sialic acid, glycoprotcins or glycolipids has been studied with particular interest, as these molecules have been postulated to contribute to various cellular receptors, and to recognition and adhesion processes (Gielen, 1968; Emmelot, 1973; Weiss, 1973).
Related biological roles have also been suggested for glycoproteins containing sialic acid on internal cellular membranes (Bosmann, 1973). However, whether sialic acid and other saccharides are present on the cytoplasmic surface of internal cellular membranes is still controversial (Hirano et al. 1972). As regards rat liver nuclear surfaces, the presence of charged groups and sialic acid has been studied with both cell-electrophoretical (e.g. Bosmann, 1973) and biochemical (e.g. Zbarsky, 1972) methods using isolated nuclei. These studies have yielded rather mixed results, probably depending partly on differences in isolation methods and experimental conditions. To provide ultrastructural confirmation of these studies, we undertook to investigate the presence of charged groups on isolated rat liver nuclei with an electron-microscope marker, colloidal iron hydroxide (CIH), known to be rather specific for groups containing sialic acid at pH values 1·7–1·8 (Benedetti & Emmelot, 1967; Weiss, Zeigel, Jung & Bross, 1972; Nicolson, 1973).
MATERIALS AND METHODS
For each experiment 2–4 rats were decapitated and their livers rapidly removed and suspended in cold physiological saline. The livers were minced with scissors in ice-cold 0–32 M sucrose solution containing 3 mM MgCl2. Homogenization was performed in 5 volumes of the same solution with a Teflon pestle homogenizer at low speed. The diluted homogenate (2×) was filtered through 2 layers of cheesecloth and centrifuged at 1000 rev/min for 10 min in an IEE model PR6 centrifuge. After centrifugation the supernatant was decanted and the pellet resuspended with a Vortex mixer in a 0·88 M sucrose solution containing 2 mM MgCL. Final purification of the nuclei was carried out with sucrose gradient ultracentrifugation. The samples were pipetted on the top of sucrose step gradient layers of 1·5, 1·8 and 2·3 M in 2 mM MgC2. In some gradients the lowermost layer was replaced with 2 M sucrose solution to increase the yield of plasma membranes. Ultracentrifugation was performed in a Spinco Model 50L centrifuge (Beckman Inc.) with a SW 25. IL rotor for 60 min at 22000 rev/min. The nuclear pellets obtained were washed several times in 0·25 M sucrose–3 mM MgCl2 solution and processed for enzyme treatments or electron microscopy.
For enzyme digestions isolated nuclei were incubated for 30 min at 30 °C in a solution containing 50 U./ml (at 37 °C) neuraminidase (Behringwerke AG, Vibrio comma) in 0·25 M sucrose–2 mM CaCL buffered with 0·01 M Tris-maleate to pH 6·0 (Drzeniek, 1973). After the treatment the samples were immersed in an ice bath and pelleted twice for 10 min at 1000 rev/min in an IEE model PR6 centrifuge at 4 °C in 0-25 M sucrose–3 m M MgCl2 solution before processing for electron microscopy.
For electron microscopy resuspended nuclei were fixed for 30 min in ice-cold 2·5% glutaraldehyde buffered with 0· 1 M sodium cacodylate to pH 7·2. Colloidal iron hydroxide (CIH) staining (Benedetti & Emmelot, 1967) was performed with carefully resuspended fixed nuclei in order to ensure good exposure to stain particles. The pH of the staining solution was adjusted to pH 1·8 with acetic acid and to pH 3·0 with NaOH. After exposure to the stain for 1 h at room temperature the samples were pelleted, fixed in 1·5% osmium tetroxide and embedded in Epon 812. Thin sections were post-stained with uranyl acetate and lead citrate or left unstained for easier visualization of CIH stain particles. Philips EM300 or Jeol 100B electron microscopes were used at an accelerating voltage of 80 kV.
Our method for isolation of rat liver nuclei was essentially a modification of the method developed by Incefy & Kappas (1971) for isolation of nuclei from chick embryo liver. The conditions of the isolation media, including the omission of buffers, were so chosen as to preserve the general ultrastructure of the nuclei and the integrity of the nuclear membranes and to retain the membrane-attached ribosomes on the outer nuclear membrane (Incefy & Kappas, 1971; Laval & Bouteille, 1973). In some experiments the yield of identifiable plasma membrane fragments was increased by modifying the gradients, to obtain an internal control for our staining method.
The isolated nuclei had a well-preserved ultrastructure (Fig. 1). Their chromatin was unaggregated and the nuclear membranes were mostly intact. Ribosomes were found in varying amounts on the outer nuclear membranes (Fig. 2 A, B).
After the fixed preparations had been stained with colloidal iron hydroxide (CIH) solution at pH 1·8, the plasma membranes left in the preparation showed dense labelling with stain particles (Figs. 3, 4). However, the nuclear surfaces seen in the same sections had only a few stain particles attached (Figs. 3, 5).
Neuraminidase treatment of the specimens (50 U./ml, 30 min at 30 °C, pH 6·0) removed most of the stain from the plasma membranes (Fig. 6) but had no apparent influence on the staining properties of the nuclear surfaces.
When stained at pH 3·0, both the plasma membranes and the outer nuclear membranes of the isolated nuclei carried heavy deposits of stain particles (Figs. 7, 8). The heavy CIH-staining pattern at pH 3·0 was not apparently altered by pretreatment of the plasma membranes and nuclei with neuraminidase (Figs. 9, 10).
Colloidal iron hydroxide staining (CIH) has been shown to be a relatively specific staining method for surface-bound N-acetyl neuraminic acid (a sialic acid) at pH values 1·7 to 1·8 (Benedetti & Emmelot, 1967; Weiss et al. 1972; Nicolson, 1973). At this pH only a few groups with low pK values can be ionized. N-acetyl neuraminic acid, with a pK value of 2·7, is partially charged (Drzeniek, 1973). In addition, it has been suggested that sulphate groups with a pK value of 1·9 (Marx et al. 1973) and the first phosphate groups of surface-bound ribonucleic acid, pK 1·0 (Weiss & Zeigel, 1972) may also contribute to the staining of membranes with CIH at pH 1·8. However, whether these 2 latter groups are present on cellular membranes is uncertain (Emmelot, 1973).
In our study the outer surfaces of rat liver cell plasma membranes showed dense labelling with CIH at pH 1·8, as also reported earlier by Benedetti & Emmelot (1967). However, no stain could be seen on the surfaces of the isolated rat liver nuclei. The lack of stain on the nuclei cannot have been due to penetration artifacts during the staining process, because the samples were thoroughly dispersed in the staining solution. In addition, in cases of appositionally situated nuclei and plasma membranes only the nuclear surfaces lacked the stain. The pK value of the first phosphate group of RNA is 1·0 (Weiss & Zeigel, 1972) and it has been suggested that ribosomal RNA is largely exposed on the ribosomal surface (Cox & Bonanou, 1969). However, at pH 1·8 we did not obtain any staining of ribosomes attached to the nuclear surface.
The disappearance of the staining of rat-liver-cell plasma membranes on treatment with neuraminidase supports the view that at pH 1·8 the attachment of CIH particles depends solely on the presence of carboxyl groups contributed by sialic acid residues. In this connexion it is interesting to note that raising the pH of the CIH staining solution to pH 3·0 causes new stainable neuraminidase-resistant groups to emerge both on the plasma membrane and on the nuclear surface. These groups may correspond to the carboxyl groups of amino acids, with a pK range of 3 to 4·0, which have been shown by cell electrophoresis to be exposed on cell surfaces at least (Vassar & Kendall, 1969). In earlier studies carboxyl groups contributed by sialic acid residues have been detected biochemically in the nuclei of rat liver cells (Kawasaki & Yamashina, 1972; Zbarsky, 1972; Phillips, 1973), L cells (Glick, Comstock, Cohen & Warren, 1971) and BHK cells (Keshgegian & Glick, 1973). On the other hand, Kashnig & Kasper (1969) reported only a negligible amount of sialic acid associated with rat liver cell nuclei. It seems that the conflicting results depend on the different techniques used in the isolation of nuclei and measurement of sialic acid (Keshgegian & Glick, 1973). At present it cannot be decided whether sialic acids are present on the surface of the outer nuclear membrane or on the cisternal surface of the nuclear envelope.
With cell electrophoresis, somewhat varying results have been obtained concerning the presence of sialic acid on rat liver nuclear surfaces (Kishimoto & Liebermann, 1964; Mayhew & Nordling, 1966; Vassar, Seaman, Dunn & Kanke, 1967; Bosmann, 1973). In a recent study Bosmann (1973) reported a marked decrease in mobility after neuraminidase treatment of rat liver cell nuclei isolated by the sucrose method. Although this study lacked ultrastructural demonstration that the nuclei were intact, the author suggested that sialic acid was present on the nuclear surface. However, in one fundamental respect the experimental conditions used by Bosmann differed from those used here: the solutions he used in electrophoresis lacked divalent cations. This probably caused detachment of the ribosomes bound to the nuclear envelope, which are known to require divalent cations for their attachment to membranes (Sabatini et al. 1972). Therefore, our observation that groups containing sialic acid are not exposed on the nuclear surfaces does not conflict with Bosmann’s electrophoretic results, if one assumes that these groups were covered by ribosomes preserved on the nuclear surfaces under our isolation conditions, including magnesium-containing media. According to Scott-Burden & Hawtrey (1973), the attachment of ribosomes to rat liver ER membranes is sensitive to neuraminidase treatment. This seems to indicate that sialic acid participates in ribosome binding to membranes. Whether under our conditions neuraminidase causes detachment of ribosomes from isolated rat liver nuclei is now being investigated.
This study was supported by grants from the Damon Runyon Memorial Foundation, the Sigrid Jus élius Foundation and the National Research Council for Medical Sciences, Finland.