Silver staining of ribosomal proteins

The procedures which are usually employed to demonstrate ribonucleoprotein in fixed intact cells depend primarily on the binding of heavy metals to the nucleic acid moiety, and, of these, the indium method of Watson & Aldridge (1961, 1964) and lead staining after aldehyde fixation (Huxley & Zubay, 1961; Reynolds, 1963; Marinozzi, 1963) have been shown to be highly selective. A number of methods have been suggested for the selective silver staining of the proteins associated with nucleic acids (Marinozzi, 1963; Black & Ansley, 1964; MacRae & Meetz, 1970) and it is evident that such methods might be helpful in elucidating some aspects of the structure of ribonucleoprotein complexes in general.

The present paper describes a procedure which appears to produce distinct and highly selective staining of the protein moiety of ribonucleoprotein complexes. Although the method also demonstrates proteins associated with DNA, this aspect is not considered in detail.

Rat and rabbit tissues were used as a matter of convenience. All tissues were fixed in 2 % glutaraldehyde in 0·1 M phosphate buffer (pH 7·3) containing 0·22 M sucrose, and some were additionally post-osmicated. After dehydration in ethanols the fixed tissues were embedded in Araldite and sectioned on a Reichert Ultramicrotome. As the staining solution which was used reacted with both copper and nickel, the sections were mounted from the boat on to titanium grids. The sections were stained by immersion of the grids in freshly prepared, unbuffered 5 % silver nitrate in double-distilled water for 30 min. The grids were removed with plastic forceps and washed by 3 immersions of 1 min each in double-distilled water.

RNA was extracted from very small pieces of glutaraldehyde-fixed rat pancreas by treatment with 10% perchloric acid at 4 °C for 24 h according to the procedures of Erickson, Sax & Ogur (1949).

The effects of DNase digestion on subsequent silver staining was assessed on separate but intact rat liver cells prepared by gentle homogenization of small pieces of rat liver in 0·1 M phosphate buffer containing 0·22 M sucrose. The homogenate was fixed by 10 % formalin in the same buffer, and, after several washings in buffer alone, was incubated for 3 h at 25 °C in a solution containing 5 mg of DNase I in 100 ml phosphate buffer containing sucrose and magnesium. A pellet of the incubated homogenate was embedded and sectioned.

Reticulocytosis was produced in rabbits of about 3 kg body weight by 4 daily sub cutaneous injections of 1 ml of 2·5 % phenylhydrazine. Two days after the last injection 2000 i.u. of heparin were administered intravenously and 20 ml of blood were removed from an ear vein. A washed pellet of cells was fixed in glutaraldehyde and embedded.

The general appearance of part of a rat pancreatic acinar cell after glutaraldehyde fixation and silver staining is shown in Fig. 1. Absence of membrane staining is evident around the nucleus and in the mitochondria, and, less obviously at this magnification, in relation to the cisternae of the endoplasmic reticulum. Although at high magnifications there is some finely granular staining within the endoplasmic reticulum (Fig. 8), the mitochondrial matrix (Fig. 10) and the nuclear sap (Fig. 3), well defined silver particles appear to be restricted to locations which are known to contain nucleoproteins, namely the nuclear chromatin, the nucleolus, the mitochondrial matrix and cytoplasmic ribosomes.

Post-osmication causes light staining of the cell membranes and also some general increase in cytoplasmic density which detracts somewhat from the clarity of the silver staining of nucleoprotein (Figs. 5, 6). Membrane definition in these circumstances appears to result entirely from osmication and not from deposition of silver as suggested by Marinozzi (1963).

The morphology of the nucleolus after glutaraldehyde fixation and silver staining is shown at different magnifications in Figs. 1‐3. The perinucleolar chromatin is sharply demarcated from both the nuclear sap and the nucleolus and is represented by a dense aggregation of silver particles which have a rather uniform size of 4 nm. The several small areas within the nucleolus which exhibit the same morphology (Fig. 3, arrows) are interpreted as strands of intranucleolar chromatin (Bernhard & Granboulin, 1968; Maggio, Siekevitz & Palade, 1963; Narayan, Muramatsu, Smetana & Busch, 1966).

The rest of the nucleolus is pervaded by fine silver granules, while, apparently within this material, there are numerous comparatively large and dispersed silver particles (Fig. 3). These 2 entities are regarded on morphological grounds as the pars fibrosa and the pars granulosa (nucleolar ribosomes) respectively. The particles of the granulosa vary in diameter from 4 to 8·5 nm with an average measurement of 6·5 nm, but in many instances (Fig. 2, long arrows) a particle of about 6·5 nm and a smaller one of about 4 nm form a closely associated pair. The total dimensions of such pairs are always less than the 15-nm diameter of complete nucleolar ribosomes (Brinkley, 1965; Hyde, Sankaranarayanan & Birnstiel, 1965; Hay, 1968), and this suggests that the paired silver particles represent 2 separate silver-staining loci within single nucleolar ribosomes. The much smaller silver granules of the pars fibrosa average 2·4 nm in diameter and in some situations (Fig. 2, short arrows) they appear to be arranged in series forming linear rows of various lengths and curvatures.

The association of cytoplasmic staining with ribonucleoprotein (RNP) is indicated by the relative appearances of the neighbouring reticulocyte and erythrocyte in Fig. 4. It is of incidental interest that in the present investigation there was no evidence of any particular concentration of RNP in relation to the reticulocyte membrane (Burka, 1968).

The chondrocyte in Fig. 5 is from cartilage which was fixed in glutaraldehyde and post-osmicated. Particulate silver staining is related to ribosomes associated with endoplasmic reticulum and to those associated with free polysomes, and at a higher magni fication of similar material (Fig. 6) it is evident that several silver particles of somewhat varying sizes are associated with each ribosome. The majority of these particles are between 4 and 6 nm in diameter and the most common number of particles per ribosome appears to be 5. Some ribosomes certainly contain fewer particles and some particles are between 6 and 10 nm in diameter, but these divergences from what is regarded as the characteristic arrangement are probably due, in part, to variations in the degree of overlap of separate silver-staining loci in variously oriented ribosomes and in part to other factors discussed later in this paper.

In the absence of post-osmication, individual ribosomes can no longer be identified, but it seems very probable that the silver particles which are related to the outer surfaces of the endoplasmic reticulum in Figs. 7 and 8 are associated with attached ribosomes in a manner similar to that noted in Fig. 6. In several situations (Fig. 8, arrows) 2 large silver particles are separated by a narrow, lucent space, and the long axis of the pair is always inclined acutely to the surface of the reticulum (Florendo, 1969). It is considered therefore that this appearance is due to the partial overlapping of multiple silver-staining loci in both the 60-s and 40-s subunits of a ribosome. In very active cells such as pancreatic acinar cells (Figs. 7, 8) cytoplasmic silver particles are not confined to the outer surfaces of the endoplasmic reticulum. Particles of the same size range, but with a higher proportion towards the lower limit of that range, lie between adjacent layers of ‘attached’ particles and sometimes form a rather distinct intermediate stratum. It is considered that this group of particles is to be associated with the ribosomes seen in the same situation after conventional staining (Palade, 1955).

In post-osmicated material (Fig. 6) mitochondrial membranes are defined, but increased electron density in the matrix tends to detract from the clarity of silve-stained structures. On the other hand, in the absence of osmication (Figs. 9, 10) the mitochondrial membranes are unstained and so the relationship of stained structures to membranes cannot be visualized.

None of the material which is stained by silver within mitochondria bears any morphological resemblance to the clumps and fibrils of DNA which are evident after nucleic acid stains (Nass & Nass, 1963a; Nass, Nass & Afzelius, 1965). However, most mitochondria do contain a considerable number of silver-stained particles which are distributed in a similar manner to the indium-stained mitochondrial ribosomes of Watson & AJdridge (1964). After post-osmication (Fig. 6) many lie close to membranes, and although others do not exhibit this relationship it cannot be excluded that they too may be similarly related to cristae lying just beyond the plane of section.

The silver-stained particles are noticeably smaller than those associated with cytoplasmic ribosomes (Fig. 10) and have a size range of 3‐7 nm with the majority between 3 and 5 nm. In a comparatively large number of instances (Fig. 10, long arrows) 2 unequal small particles lie close together, the total dimensions of the pair being less than those of negatively stained mitochondrial ribosomes (O’Brien & Kalf, 1967). Occasionally (Fig. 10, short arrows) 3 or 4 particles form a row, suggesting that 2 of the pairs noted above may be closely associated.

It has been found that although particulate silver staining is confined to regions containing nucleic acids, procedures which remove nucleic acids affect the staining in only one of these situations.

Incubation of a suspension of formalin-fixed rat liver cells with DNase causes no significant alteration in the staining of nuclear chromatin (Fig. 13).

The removal of RNA by treatment with cold perchloric acid (PCA) appears to have no effect on the silver staining of cytoplasmic ribosomes (Figs. 11, 12). Neither does this procedure abolish the silver particles in mitochondria, though these are partially obscured by an increase in the electron density of the mitochondrial matrix sufficient to negatively outline the cristae (Fig. 11). A similar increase in density after PCA treatment was noted by Watson & AJdridge (1964), who ascribed it to complexing of retained PCA with indium. In contrast to these findings in the cytoplasm and mito chondria, cold PCA treatment causes a total abolition of silver staining in the pars fibrosa and pars granulosa of nucleoli (Fig. 12).

In the material examined, silver-stained particles were observed only in relation to structures consisting of nucleic acids and associated proteins. Although silver ions bind to both DNA and RNA in solution (Jensen & Davidson, 1966; Daune, Dekker & Schachman, 1966; Ivanova, Minchenkova & Timofeeva, 1967) it is improbable that this is the basis of the staining reaction in fixed tissues. Thus although incubation of formalin-fixed tissues with DNase abolishes the nuclear Feulgen reaction and prevents uranyl-acetate staining of DNA (Ris & Plaut, 1962; Nass & Nass, 1963 b; Kislev, Swift & Bogorad, 1965) it appears to have no effect on the silver staining of chromatin. Similarly cold perchloric acid treatment, which removes 85 % of the total RNA (Watson & Aldridge, 1964; Savitsky & Stand, 1965) in contrast to the 20‐30% of ribosomal RNA removed by incubation with RNase (Cox, 1969), and which extracts only 3‐4% of ribosomal proteins (Horowitz & Schechter, 1968), does not affect silver staining of cytoplasmic or mitochondrial ribosomes. These observations strongly suggest that silver staining depends on the binding of silver to nucleic acid-associated proteins: Marinozzi (1963) and Black & Ansley (1964) using other silver methods came to similar conclusions. This interpretation is supported by the failure of the method to demonstrate mitochondrial DNA, which is probably not associated with histones in the manner characteristic of DNA in eukaryotic nuclei (Nass et al. 1965), and although it does not appear to be in keeping with the abolition of nucleolar silver staining by cold perchloric acid treatment, it may be that that effect is due to a peculiarity, of maturity rather than kind, of the nucleic acid-associated proteins in this situation. The mechanism of this silver staining of proteins is not yet understood but it is possible that the reaction depends on silver ions forming chelate complexes with the peptide backbone, and that the observed selectivity for nucleic acid-associated proteins may depend on a particular protein conformation which satisfies the steric requirements of the reaction (Steinhardt & Beychok, 1964).

There is considerable evidence (Hay, 1968) which indicates that the pars fibrosa of the nucleolus consists of a fibrillar form of RNP having extended 45-s RNA as its nucleic acid moiety, and it is suggested that the linear series of 2·5-nm silver-stained particles observed in this region (Fig. 2) represent loci of protein distributed along this form of RNA. Although the pars granulosa contains a number of species of RNP particles, it is probable that the majority of those present at any one time are 62-s nucleolar ribosomes (Perry, 1966; Penman, 1966; Liau & Perry, 1969). The dimensions and relationship of the pairs of silver-stained particles observed as the predominant feature of this region suggest that they represent 2 unequal protein loci in these ribosomes. Moreover, since 62-s nucleolar ribosomes eventually migrate and become the 60-s subunits of cytoplasmic ribosomes, with little change in either the absolute or proportional amounts of their protein content (Liau & Perry, 1969), it seems reason able to assume that the cytoplasmic 60-s subunit has an essentially similar morphology.

It has been shown that cytoplasmic ribosomes usually contain 5 silver-stained particles. If, as suggested above, the 60-s subunit contains 2, it follows that the other 3 must represent protein loci in the 40-s subunit, and this appears to be in keeping with the higher protein/RNA ratio of the smaller subunit - 1·4 as against 0·76 (Perry & Kelley, 1968)- and with the tripartite structure noted in some small subunits in chloroplast ribosomes (Miller, Karlsson & Boardman, 1966). Variation in the sizes of the 5 silver-stained protein loci in cytoplasmic ribosomes may reflect the very considerable differences which have been shown to exist in the molecular weights of individual ribosomal proteins (Moore, Traut, Noller, Pearson & Delius, 1968; Craven, Voynow, Hardy & Kurland, 1969), and, possibly, the existence of different kinds of proteins in different members of a ribosome population (Craven et al. 1969). A cytoplasmic ribosome structure involving the presence of 5 protein loci 4‐6 nm in size appears to be compatible with X-ray diffraction studies (Langridge & Holmes, 1962; Langridge, 1963) which indicated that these ribosomes contained 4 or 5 RNA double helices separated by intervals, presumably occupied by protein, of 4·5—5·5 nm.

The pairs of small silver-stained particles noted in the matrix of mitochondria are regarded as 2 protein loci in mitochondrial ribosomes. Interpreted in this way the distribution of the particles is similar to that of mitochondrial ribosomes demonstrated by nucleic acid stains (Watson & Aldridge, 1964; Rabinowitz et al. 1965; Luck, 1964). The occasional association of 3 or 4 silver particles in a row may represent the association of 2 mitochondrial ribosomes as a polysome-like dimer as suggested by Perlman & Penman (1970).