Ultrastructural distribution of DNA and RNA within the nucleolus of human Sertoli cells as seen by molecular immunocytochemistry

On the basis of autoradiographic studies, it used to be accepted that rDNA transcription takes place in the dense fibrillar component of the nucleolus (reviewed by Goessens, 1984; Fakan, 1986). Over the past few years, however, data based on immunolocating procedures at the electron micro-scope level have strongly suggested that rDNA transcrip-tion occurs in the fibrillar centres (Scheer and Benavente, 1990; Thiry et al., 1991). A variety of methods, applied to various cell-type nucleoli, have detected DNA exclusively in the fibrillar centres, preferentially in their peripheral regions, but not in the dense fibrillar component (Scheer et al., 1987; Thiry, 1988, 1992; Thiry et al., 1988, 1991, 1992).

Recent in situ hybridization studies have, however, yielded apparently contradictory results concerning the loca-tion of rRNA genes in the nucleolar fibrillar components. In Ehrlich tumour cells (Thiry and Thiry-Blaise, 1989, 1991), HeLa cells, and mouse 3T3 cells (Puvion-Dutilleul et al., 1991), rDNA was detected only in the fibrillar centres and not in the dense fibrillar component, while the reverse was observed for human Sertoli cells (Wachtler et al., 1992) and human lymphocytes (Wachtler et al., 1990).

Although the functional organization of the nucleoli may be different in different cells or species, we have put for-ward the hypothesis that the divergent conclusions might arise from difficulties in distinguishing the various nucleolar components under in situ hybridization conditions (Thiry and Goessens, 1992). In this respect, we have recently shown that the rDNA signal obtained after in situ hybridization on human lymphocytes should be attributed to the condensed chromatin crossing through the dense fib-rillar component rather than to the dense fibrillar compo-nent itself (Vandelaer et al., 1992).

To shed light on the nucleolus of human Sertoli cells, we have used the in situ terminal deoxynucleotidyl trans-ferase/immunogold (TdT/immunogold) technique to inves-tigate, in great detail, the precise distribution of DNA within the nucleolus. This method was applied to acetylated mate-rial, conditions that provide an excellent distinction between the various nucleolar components and make it possible to visualize condensed chromatin with high contrast. To obtain additional information concerning the morphofunctional organization of the nucleolus, the distribution of RNA was further investigated by means of the in situ polyadenylate nucleotidyl transferase/immunogold (PnT/immunogold) technique.

The results reveal DNA in the chromatin enclosed in the nucleolar interstices, especially those in contact with the fibrillar centres, as well as in the fibrillar centres, but not in the dense fibrillar component of the human Sertoli cell nucleolus. Moreover, our results indicate that the fibrillar centre is the only site where DNA and RNA are visualized together.

Testicular material from human adults was obtained by courtesy of Dr P. Delrée (University of Liège, Liège, Belgium).

Small fragments of testicular material were fixed for 60 min at 4°C in 1.6% glutaraldehyde in 0.1 M Sorensen’s buffer (pH 7.4), acetylated as previously described (Wassef et al., 1979), and embedded in Epon. Ultrathin sections were mounted on nickel grids and stained with uranyl acetate and lead citrate before exam-ination in a Jeol CX 100 electron microscope at 60 kV.

To pinpoint the precise location of DNA and RNA, respectively, the sensitive TdT/immunogold technique (Thiry, 1992a,b) and the sensitive PnT/immunogold procedure (Thiry, 1992c) and their controls were used as previously described.

To evaluate the labelling density, the area of each compartment studied was first estimated morphometrically by the point-count-ing method (Weibel, 1969). After evaluating the areas (S_a) occu-pied by the various compartments, we counted the number of gold particles (Ni) over each compartment and calculated the labelling density (Ns = Ni/Sa). In the present study, to evaluate the DNA and RNA labeling densities, 21 and 7 random micrographs were analyzed, respectively and 5,645 and 2,986 gold particles counted, respectively. We must bear in mind that these numerical data do not reflect the exact amounts of nucleic acid molecules in the com-partments studied; thus, only relative intensities between com-partments can be considered. This quantitative evaluation allows only a demonstration of the specificity of the labelling.

On electron micrographs, the human Sertoli cell nucleus is easily recognized by the following criteria. The chromatin appears uniform, distributed throughout the nucleus, and essentially constituted of fibres approximately 25 nm in diameter. The nuclear envelope shows many deep invagi-nations. A single nucleolus stands out very clearly and shows a typical arrangement.

Three distinct parts form the nucleolus (Fig. 1A,B): one or sometimes two large fibrillar centres, a granular mass, and strands containing dense fibrillar component and gran-ular component. These strands connect the fibrillar centres with the granular mass. Only those parts of the strands that contain dense fibrillar component come in contact with the fibrillar centres. Apart from these connections, the fibrillar centres appear separated from the other nucleolar compo-nents by nucleolar interstices. Numerous nucleolar inter-stices are also found between the strands themselves, and between the strands and the granular mass. Inside these spaces, chromatin fibres or a few rare, small clumps of con-densed chromatin are consistently observed. This chromatin can be visualized in close contact with the fibrillar centres.

When the in situ TdT/immunogold technique is applied to acetylated cells, gold particles are found scattered through-out the nucleoplasm (Fig. 2A). Only the interchromatin granules appear free of gold particles.

Inside the nucleoli (Fig. 2A,B), label is preferentially seen over the nucleolar interstices, notably those bordering on the fibrillar centres. Evident label is also consistently present over the fibrillar centres, centrally as much as peripherally. In contrast, although gold particles are fre-quently observed at junctions between the strands and the intersitices, no label is visualized over the fibrillogranular strands themselves. The granular masses are also devoid of label.

Numerical data on gold particle distribution over the var-ious nucleolar components are summarized in Table 1. They confirm the subjective observations and further establish the high specificity of the labelling. In fact, the labelling den-sity over the nucleolar interstices approaches the value measured over the nucleoplasm. Although much lower than that obtained over the nucleolar interstices, significant label is also found over the fibrillar centres. Over the dense fib-rillar component and the granular component, on the other hand, the labelling density is insignificant.

To pinpoint the location of RNA within the nucleolus, the in situ PnT/immunogold procedure was applied to acety-lated cells. Under these experimental conditions, both cyto-plasmic and nuclear compartments of interphase cells are labelled. In the nuclei, label is particularly obvious over the nucleoli (Fig. 3). Inside all the nucleoli, evident labelling is present over the dense fibrillar component and the gran-ular component. In the fibrillogranular strands, gold parti-cles are consistently visualized both centrally and periph-erally. Label is also detected over all the fibrillar centres. Very faint labelling is seen over some nucleolar interstices. When a nucleolar interstice contains a small clump of con-densed chromatin, the latter is never labelled.

Numerical data on gold particle distribution over the var-ious nucleolar components are summarized in Table 2. They confirm the subjective observations and further establish the high specificity of the labelling as illustrated by the absence of gold particles over the rare clumps of condensed chro-matin. In fact, the labelling density over the fibrillar cen-ters, although higher than that obtained over the granular component, has a value near to the labelling density over the dense fibrillar component of the nucleolus. Over the nucleolar interstices, on the other hand, the labelling den-sity is clearly lower. The labelling density over the nucle-oplasm has a value similar to that measured over the gran-ular component.

The present study shows that DNA is present in the nucle-olar interstices and the fibrillar centres of human Sertoli cell nucleoli. As for the dense fibrillar component, it appears to be completely devoid of DNA.

These results appear to contradict the recent data obtained on the same cell type by in situ hybridization (Wachtler et al., 1992). In fact, the in situ hybridization signal was attributed to the dense fibrillar component because these authors considered that the strands were entirely composed of dense fibrillar component in the nucle-oli of human Sertoli cells. However, as shown here and in previous studies (Devictor et al., 1984, 1987) the strands are not exclusively formed by dense fibrillar component but comprise alternating segments of dense fibrillar component and granular component. On the other hand, these strands take on a reticular appearance due to the presence of numer-ous interstices that the present work shows to contain chro-matin. Although Wachtler et al. (1992) frequently observed label at the periphery of these strands and, thereby, as shown here, at the periphery of the nucleolar interstices, they took no account of the nucleolar interstices, perhaps because they were hard to identify under in situ hybridiz-ation conditions.

Accordingly we think that the rDNA signal obtained after in situ hybridization on human Sertoli cell nucleoli should be attributed to the chromatin enclosed in the nucleolar interstices rather than to the dense fibrillar component itself. In this regard, it is pertinent to note that, apart from a few cytochemical studies suggesting the presence of DNA inside the dense fibrillar component, DNA has never been clearly demonstrated there (discussed by Thiry and Goessens, 1992). Moreover, the presence of rRNA genes in the intranucleolar chromatin enclosed in interstices is in complete agreement with previous data obtained on a vari-ety of cell types in both biochemical (Bachellerie et al., 1977) and in situ hybridization studies (Jacob et al., 1974; Thiry and Thiry-Blaise, 1989, 1991; Puvion-Dutilleul et al., 1991, 1992; Stahl et al., 1991). In Ehrlich tumour cell nucle-oli, we have even previously pointed out the presence of rDNA in the condensed chromatin interrupting the dense fibrillar component and in contact with the fibrillar centres (Thiry and Thiry-Blaise, 1989, 1991).

These results extend our previous observations obtained on other cell types, concerning the distribution of DNA within the nucleolus (Thiry, 1992a); they also emphasize the fact that it is important to know the fine distribution of DNA within the nucleolus before investigating the distrib-ution of more specific DNA segments associated with the nucleolus.

The present study also shows that RNA is present in the fibrillar centres of the human Sertoli cell nucleolus but not in nucleolar interstices containing chromatin. The presence of RNA in the fibrillar centres is in complete agreement with recent data obtained on a few other cell types by the PnT/immunogold technique, and by postembedding immunogold labelling procedures using two different anti-RNA antibodies (Thiry, 1992b). In Ehrlich tumour cell nucleoli, moreover, we have recently demonstrated the presence of appreciable amounts of rRNA in the fibrillar centres by electron microscope in situ hybridization (Thiry, 1992c).

The fibrillar centres of the human Sertoli cell nucleolus appear, as in other cell types, to be the only nucleolar com-ponent in which significant amounts of both nucleic acids are visualized together. These results are in line with the view that rRNA synthesis takes place in the fibrillar cen-tres of mammalian cell nucleoli and indicate further that the dense fibrillar component cannot be considered as formed by superposition of active rRNA genes and their transcripts, but should rather be seen as a component where primary transcripts are rapidly accumulated after their release from the rDNA template (Scheer and Benavente, 1990; Thiry et al., 1991).

The author is grateful to Prof. G. Goessens (University of Liège, Belgium) for encouraging discussions and for critical reading of the manuscript. He also acknowledges the skilful technical assistance provided by Miss F. Skivée and J.-C. Paulus. This work received financial support from the ’Fonds de la Recherche Scientifique Médicale’ (grant no. 3.4512.90) and from the ’Actions de Recherche Concertées’ (grant no. 91/95-152).