Electron microscopic evidence for RNA polymerase loading at repeated sequences in non-transcribed spacers of D. virilis.

The organization of transcribed and non-transcribed sequences in transcriptionally active ribosomal DNA (rDNA) chromatin from Drosophila virilis nurse cells (excluding follicle cells) was examined by electron microscopy. Nuclear spread preparations revealed that the actively transcribing rRNA genes (rTUs) exhibited a single uniform length distribution (7 +/- 0.5 kb). Non-transcribed 'spacer' regions separating active rTUs were, on the other hand, quite heterogeneous in length with a mean of 1.95 +/- 1.4 kilobase pairs (kb). Within each 'spacer' segment of chromatin two morphologically distinct regions were discernable: (1) A fiber-free region adjacent to the 3' termination site of the preceding rTU which accounted for the majority of each non-transcribed 'spacer'. This region was highly heterogeneous in length, and was termed the 'Non-repetitive Spacer' (NRS). (2) A short region, largely uniform in length, which was located immediately 5' to each rTU. This region was covered by a linear array of RNA polymerase molecules with a few very short fibers. The arrangement of polymerase in these arrays and their proximity to rRNA coding sequences appeared to be morphologically indicative of multiple initiation sites which are thought to be present in this region of the non-transcribed spacer as evidenced by direct sequence analysis (Coen, E S & Dover, G A, Nucl acid res 10 (1982) 1017 and Miller, J R et al., Nucl acid res 11 (1983) 11) [6, 18]. The data suggest that the repeated sequences in the non-transcribed spacer region serve as multiple in vivo loading and/or initiation sites for RNA polymerase I. Our results further indicate that, although an estimated 75% of the ribosomal RNA genes in Drosophila virilis are interrupted by an intron in the 28Sb gene sequence (Barnett, T & Rae, P M M, Cell 16 (1979) 763) [1], none of those genes appear to be transcribed to any appreciable extent in the tissues examined. It is also reasonable to infer that the actively transcribing uninterrupted genes (and therefore the intron-containing genes as well) are clustered and not randomly interspersed.