The snoRNAs represent an abundant group of non-protein-encoding RNAs that function in the biogenesis of rRNA and snRNAs (reviewed in reference 15
). As most classes of cellular RNAs, mature snoRNAs are generated by a complex processing pathway. In strains that are defective for the exosome, 3′-extended forms of many snoRNAs are detected, although it is currently unclear to what extent these represent maturation intermediates or RNAs that have been identified by surveillance activities and targeted for degradation by the exosome (1
). Many of these extended snoRNAs are polyadenylated, suggesting that the addition of the poly(A) tails would normally promote their degradation (18
). This implies that a significant fraction of the snoRNA precursor population may normally be degraded by the exosome. Here we report that in yeast cells lacking the exosome-associated exonuclease Rrp6, polyadenylated RNAs, including 3′-extended and polyadenylated forms of the U14 snoRNA, accumulate in a discrete compartment within the nucleolus.
Because most eukaryotic mRNAs have a 3′ poly(A) tail, the detection of poly(A)+
RNA by in situ hybridization with oligo(dT) probes has generally been considered to reflect the distribution of cellular mRNA. Our data for the wild-type yeast strains were indeed consistent with this interpretation. Poly(A) staining was detected predominantly in the cytoplasm and was greatly reduced in a strain carrying a mutation in Pap1, the polymerase responsible for the synthesis of poly(A) tails on nuclear pre-mRNAs (21
). Poly(A) staining was also observed in the nucleoplasm, albeit at lower levels than in the cytoplasm, but the staining was not detected in the nucleolus (Fig. ). This indicates that polyadenylated mRNAs are detectable in the nucleoplasm, whereas the nucleolus contains little polyadenylated RNA in a wild-type cell. We observed a region of higher poly(A) staining intensity in close proximity to the nuclear rim, which was identified by a GFP-tagged nucleoporin, possibly reflecting a local concentration of mRNAs in association with the nucleocytoplasmic transport machinery.
A substantially different pattern of poly(A) staining was observed in a strain lacking Rrp6. The poly(A)+
RNA signal was greatly increased in the nucleus (Fig. ) and was particularly enriched in a subnucleolar focus, which we refer to as the nucleolar poly(A) domain (Fig. ). Microarray analyses have not revealed dramatic increases in the levels of most mRNAs in the rrp6
Δ strain at 37°C (10
), and visualization of the ACT1
mRNA in the Rrp6-depleted strain revealed no accumulation in the nucleus or in the nucleolus (Fig. ). This result indicates that the polyadenylated RNA species that are being detected in the nucleolar poly(A) domain are unlikely to correspond to mRNAs.
As shown in Fig. , the distributions of the U14 snoRNA and the snoRNA-associated protein Nop1 were drastically altered in the rrp6
Δ strain at the nonpermissive temperature. Nop1 and U14 showed a uniform nucleolar staining at 23°C but became highly enriched in a focus that precisely colocalized with the nucleolar poly(A) domain following transfer to 37°C. Since strains lacking Rrp6 accumulate 3′-extended and polyadenylated snoRNAs, including U14 (Fig. and ), our results strongly suggest that the nucleolar poly(A) domain contains polyadenylated snoRNAs. This idea is supported by the finding that U3 snoRNA, which was not found to be polyadenylated in rrp6
Δ strains (1
), appeared to be excluded from the nucleolar poly(A) domain (Fig. ). The maturation of snoRNPs, including U14, involves passage through an NB that may be related to the human Cajal body (37
). However, an analysis of the localization of the snoRNA cap-trimethylase Tgs1 showed that the nucleolar poly(A) domain and NB are physically distinct (Fig. ).
The use of a probe specific for the 3′-extended form of U14 snoRNA showed that the accumulated poly(A)+
forms are indeed present in the poly(A) domain. However, the Northern analyses show that these represent only a small fraction of the total U14 population, indicating that mature, presumably functional U14 can also localize to the poly(A) domain. Analyses of strains carrying point mutations in Sda1 identified a subnucleolar structure termed the No-body (7
). Sda1 is associated with a late pre-60S ribosomal subunit and is required for its export to the cytoplasm. Following the inhibition of 60S ribosomal subunit export in an sda1
mutant strain, the NB is very strongly and rapidly enriched for pre-60S particles, which are substrates for polyadenylation by Trf4 and degradation by the exosome. However, pre-40S particles that are not known to be defective in sda1
mutants are also accumulated, presumably transiently, in the NBs. This result suggests that the production of abundant degradation substrates induces the formation of visible surveillance centers. These detectably contain both the defective RNA-protein complexes and at least some “normal” complexes that will presumably pass the surveillance steps and be released. The relationship between the NB and the nucleolar poly(A) domain reported here remains to be determined. It is, however, unlikely that they are simply identical, since the loss of Rrp6 promoted the formation of the nucleolar poly(A) domain but inhibited NB formation (7
The degradation of nuclear RNAs by the exosome is activated by the TRAMP complex (18
), and our results show that TRAMP is required for the formation of the nucleolar poly(A) domain (Fig. and ). The poly(A) polymerase activity of Trf4 is distinct from that of Pap1, which is responsible for the synthesis of poly(A) tails of nuclear mRNA precursors. Within the mRNA 3′ synthesis machinery, Pap1 is highly processive and rapidly adds 60 to 90 adenine residues to mRNAs. In contrast, Trf4 exhibits a slow and distributive polyadenylation activity in vitro (18
). In wild-type cells, the poly(A) tails synthesized by Trf4 may never normally be extended to more than a few nucleotides, due to rapid degradation by the exosome (18
), whereas in situ detection by the (dT)50
probe presumably requires the presence of long A tails. In the absence of exosome activity, the short oligo(A) tails may act as primers for polyadenylation by Pap1 (18
). Consistent with this model, reduced polyadenylation of both snoRNA and rRNA precursors was reported in a pap1
Δ strain relative to the rrp6
Δ single mutant (17
). This result may explain the great reduction in the nucleolar poly(A) domain seen in the pap1
Δ double-mutant strain at 37°C (Fig. ). However, it is also possible that the effects of pap1
mutations are indirect consequences of defective mRNA polyadenylation and subsequent mRNA instability.
Rrp6 and Mtr4 are implicated in the surveillance and degradation of many different nuclear RNA species (5
). The lack of nucleolar poly(A) staining in normal cells is consistent with the view that polyadenylated RNAs are normally very short-lived intermediates that are rapidly degraded by the exosome. The absence of Rrp6 results in the accumulation of polyadenylated forms of snoRNAs, snRNAs, pre-rRNA, rRNA, intergenic cryptic unstable transcripts, and other transcripts, each of which presumably contributes to the strong nuclear accumulation of poly(A)+
RNAs. It remains to be established whether all of these species are enriched in the nucleolar poly(A) domain. Since an exosome component, Rrp43, is also detected to be enriched in the poly(A) domain (Fig. ), it is most likely that this domain corresponds to a dedicated compartment where polyadenylation and degradation of nucleolar RNAs takes place. We speculate that the compartmentalization of polyadenylated RNAs may promote their efficient recognition as substrates for degradation by the exosome.