Early transfer RNA (tRNA) processing events in Saccharomyces cerevisiae are coordinated in the nucleolus, the site normally associated with ribosome biosynthesis. To test whether spatial organization of the tRNA pathway begins with nucleolar clustering of the genes, we have probed the subnuclear location of five different tRNA gene families. The results show that tRNA genes, though dispersed in the linear genome, colocalize with 5S ribosomal DNA and U14 small nucleolar RNA at the nucleolus. Nucleolar localization requires tRNA gene transcription-complex formation, because inactivation of the promoter at a single locus removes its nucleolar association. This organization of tRNA genes must profoundly affect the spatial packaging of the genome and raises the question of whether gene types might be coordinated in three dimensions to regulate transcription.
Little is known about how specific nuclear genes are arranged in three dimensions, although the position of some chromosome regions are dynamic and responsive to the general transcription state (1). Examples of documented DNA positions are the telomeric regions, preferentially found near the nuclear periphery in yeast, and the tandemly repeated ribosomal RNA (rRNA) genes, which create distinctive nucleolar structures where ribosome assembly is coordinated.
Components of the early tRNA processing pathway in Saccharomyces cerevisiae and other eukaryotes are often at the nucleolus (2–5), and additional observations suggest that transcription of the tRNA genes might also be nucleolar. Transcription by RNA polymerase II (Pol II) is suppressed in the vicinity of tRNA genes (6, 7), and promoters for Pol II are underrepresented within 500 base pairs upstream of tRNA genes, with the exception of Ty retrotransposons (8, 9). This silencing near tRNA genes is relieved by a mutation in the rRNA processing enzyme, CBF5 (10). These observations suggest that tRNA and rRNA biogenesis might be coordinated at the nucleolus and that spatial coordination might begin by localizing the tRNA genes at the nucleolus (10). Such a clustering of tRNA genes would create a region that was highly concentrated for Pol III as well as Pol I transcription. In yeast, the Pol III–transcribed 5S rRNA genes are nucleolar by definition, because they adjoin the large rRNA gene repeats (11). Nucleolar localization of 5S rRNA genes has also been found in higher eukaryotes (12–16), in which the 5S gene clusters are not attached to the large rRNA genes in the linear DNA.
We investigated the location of five different families of tRNA genes: tRNALeu(CAA) and tRNALys(CUU), which contain introns, and tRNAGly(GCC), tRNAGln(UUG), and tRNAGlu(UUC), which do not contain introns. Each family contains 9 to 16 genes at scattered positions throughout the yeast genome (fig. S1). Fluorescent oligonucleotide probes specific for the antisense strand of the genes were hybridized to fixed, permeabilized yeast, and the signal was compared to that obtained with probes to either a small nucleolar RNA (U14 snoRNA) or a known nucleolar DNA segment from chromosome XII, the 5S rRNA gene (17).
Most of the signal from all five families of tRNA genes overlaps the 5S ribosomal DNA (rDNA) at the nucleolus (Fig. 1). The tRNA genes tend to be on the periphery of the 5S rDNA signal, but it is not clear whether they are within or at the edge of the nucleolus (5S rDNA and U14 snoRNA positions). tRNA gene probes were specific for DNA, because the signal was sensitive to deoxyribonuclease but not ribonuclease or proteinase (fig. S2). We are probably able to detect only clustered tRNA genes; individual genes would likely fall below the detection limit. Consistent with this, some pre-tRNAs are nucleoplasmic, although most of the pre-tRNA signal is nucleolar (2, 8, 18, 19). The nucleolar signal is specific to tRNA genes, because telomeric probes give a signal in the nucleoplasm and nuclear periphery, as expected (20).
Nucleolar localization of tRNA genes. Fluorescent oligonucleotide probes were annealed to individual tRNA gene families (Leu, Gly, Gln, Lys, and Glu) or to the telomeric repeats. Each family of tRNA genes has 9 to 16 members that are dispersed in the linear genome map (28) (fig. S1). The Cy3-labeled probes to the tRNA genes and telomeres are red, and the Oregon Green 488 probe to the 5S rRNA gene is green. 4′,6′-diamidino-2-phenylindole (DAPI) staining of the nucleoplasm is shown as blue. Most of the tRNA gene signal overlaps the nucleolar 5S rRNA genes. In contrast, the telomeric repeat probe stains the nuclear periphery and nucleoplasm. Cells shown are representative of multiple experiments.
To test whether tRNA gene position is dependent on Pol III transcription, we compared the position of a single tRNA gene locus, SUP53 on chromosome III (Fig. 2), to the position of the same locus after the SUP53 promoter had been inactivated by point mutations that precluded stable association of all Pol III transcription-complex components (21). When the tRNA gene is active, the SUP53 signal overlaps the nucleolar signal ~50% of the time, which is substantially more frequent than the signal overlap that occurs when the tRNA gene is inactive (52% of 440 cells compared with 13% of 715 cells). Similar distributions of the locus were found when the same active and inactive tRNA genes were placed with URA3 on a low-copy plasmid (7) (fig. S3). These results suggest that a given tRNA gene might not necessarily be found continuously at the nucleolus but that the tendency for nucleolar association is dependent on some aspect of transcription-complex formation.
tRNA gene localization depends on Pol III complex formation. The SUP53 tRNA gene locus on chromosome III was probed by replacing the neighboring LEU2 coding region with URA3 coding sequences and simultaneously probing fixed cells with 14 fluorescent oligonucleotides (red) complementary to the non-RNA strand of URA3. Nucleoli were identified with a fluorescent oligonucleotide to U14 snoRNA (green), and the nucleoplasm is stained with DAPI (blue). The endogenous URA3 coding sequences are deleted from chromosome V in these strains. As a negative control, a strain in which the URA3 sequences were deleted by replacement with a kanamycin resistance gene (Kan) showed no signal with the URA3 probes (lowest panel). SUP53 frequently localized in or showed some overlap with the nucleolus when transcriptionally active (52% out of 440 cells imaged), but did so less often when the promoter was mutated to be inactive (13% out of 715 cells imaged).
To test whether tRNA gene clustering is sensitive to nucleolar perturbation, we probed the tRNALeu family in cells with a nonlethal deletion of the Pol I subunit, Δrpa49 (Fig. 3), which compromises Pol I transcription and results in a less compact nucleolus than normal (22) (Fig. 3). In keeping with results from previous nucleolar disruption (10), the pre-tRNA transcript signal in Δrpa49 disperses into the nucleoplasm. The tRNA gene signal disappears, because the dispersed tRNA genes no longer individually produce enough of a signal to be seen above the background fluorescence. We conclude that tRNA gene localization is, to some extent, dependent on normal rRNA biosynthesis in the nucleolus.
Pol I affects tRNA gene clusters. The positions (red) of the tRNALeu gene family and its pre-tRNA transcripts were probed (17) in a strain with a deleted, nonessential subunit of Pol I (Δrpa49) and compared with the parental wild-type strain (WT). Nucleoli are marked by a probe to the U14 snoRNA (green) and nucleoplasm is stained with DAPI (blue) in all panels.
Clustering tRNA genes near 5S rRNA genes in the nucleolus forms a nuclear subregion specializing in Pol III transcription. Transcription complexes on 5S rRNA genes and tRNA genes both contain Pol III and two multisubunit transcription factors, TFIIIB and TFIIIC, although the 5S rRNA genes require one extra recognition factor, TFIIIA (23). Spatial coordination of the Pol III–transcribed genes would allow concentration of these factors, as well as initiating nucleolar organization of the early tRNA processing pathway (1, 9).
Localization of most tRNA genes to the yeast nucleolus has drastic implications for three-dimensional genome organization, because the genes are scattered throughout the linear genome map (fig. S1). It is not clear whether these results reflect the organization of tRNA genes in metazoans. There is currently little information on the position of actively transcribed tRNA-class promoters in metazoans, including both tRNA genes and short interspersed nuclear elements (24). Further, morphology of nucleoli in multicellular eukaryotes is distinct from the single nucleolus in yeast.
The clustering of tRNA genes in subnuclear regions specific for Pol III expression (or Pol I and Pol III expression) is consistent with Pol II transcription being relatively poor near actively transcribed tRNA genes (7, 10) and with Pol II promoters being underrepresented near tRNA genes (9). A case of transcription-unit segregation, the rRNA genes in nucleoli, results in the exclusion of Pol II from nucleoli and the silencing of Pol II transcription units when they are inserted between rRNA gene repeats (25). If Pol III transcription units are also localizing in or near the nucleolus, Pol II transcription might be suppressed by either simple exclusion of transcription components or a more active mechanism. This clustering of tRNA genes in specialized subnuclear locations might also partially explain why propagating a chromatin-modeling signal beyond a tRNA gene is difficult, as observed in cases where tRNA genes serve as boundary elements for chromatin domains (26, 27).