The processes and trans
-acting factors involved in the regulation of H/ACA snoRNP assembly, intranuclear trafficking, and activity remain largely unknown. We are particularly interested in these issues. Our attention was therefore drawn to the Naf1p factor when it was reported that it interacts in a double-hybrid test with two core components of H/ACA snoRNPs, Nhp2p and Cbf5p (32
). In accordance with those results, we find by immunoprecipitation experiments that Naf1p interacts in cell extracts, directly or indirectly, with several components of H/ACA snoRNPs. We believe that the interaction is specific because we fail to detect any significant immunoprecipitation with epitope-tagged Naf1p of components of C/D snoRNPs or of the control Nsr1p protein. Our conclusion that Naf1p physically associates with H/ACA snoRNP components is reinforced by the recent finding that Naf1p coprecipitates with Cbf5p tagged with the Flag epitope (30
). The efficiencies of precipitation of H/ACA snoRNP components with Naf1p are clearly low, which is probably indicative of a transient and/or labile association.
Naf1p is found mainly in the nucleoplasm, cannot be detected in the dense fibrillar component of the nucleolus, and sediments in a glycerol gradient above all as a free protein or as part of a complex of the size of free snRNPs. This suggests strongly that (i) Naf1p is not a core component of active mature H/ACA snoRNPs, consistent with a previous proposal that most mature H/ACA snoRNPs contain only the Cbf5p, Gar1p, Nhp2p, and Nop10p proteins (29
) and that (ii) Naf1p is not directly involved in pre-rRNA processing and modification. We are fully aware that the Naf1p presence in the dense fibrillar component of the nucleolus may have gone undetected, and we observe that traces of Naf1p sediment with preribosomes. Nevertheless, the likeliest interpretation of the 18S rRNA accumulation defect detected in Naf1p-depleted cells remains that it is the consequence of the reduction in snR10 and snR30 H/ACA snoRNP levels. Indeed, the most striking phenotype resulting from Naf1p depletion is a dramatic and specific decrease in the accumulation of all H/ACA snoRNP components except Nhp2p. Remarkably, none of the other small RNAs tested, in particular none of the C/D snoRNAs, show a similar decrease in steady-state levels. The very specific reduction in H/ACA snoRNA levels following Naf1p depletion was also demonstrated by the groups of G. Chanfreau and D. Tollervey (P. K. Yang, G. Rotondo, T. Porras, P. Legrain, and G. Chanfreau, submitted for publication; and A. Fatica, M. Dlaki, and D. Tollervey, submitted for publication). This accumulation defect is probably not due to defective transcription, since independently transcribed H/ACA snoRNAs and the snR44 intronic RNA are equally affected. In the case of snR44, our results strongly suggest that no defective transcription (either initiation, elongation, or termination) is occurring: snR44 production has been shown to be dependent upon splicing of its host pre-mRNA (65
), and we failed to detect a significant decrease in the steady-state accumulation of the corresponding mRNA. Defective pre-snoRNA processing in Naf1p-depleted cells is not very probable either. The activity of the enzymatic machinery (i.e., splicing machinery, Rnt1p endonuclease, Rat1p exonuclease, and exosome) as such is not impaired, since it is involved in processing precursors to both H/ACA and C/D snoRNAs. Yet no defective processing of C/D snoRNAs is detected. For example, we notice that the pattern of bands detected with the U24 or snR78 probes reflecting heterogeneous end formation by exonucleases is not altered by Naf1p depletion. We cannot of course exclude the possibility that the processes that target the processing machinery to H/ACA snoRNA precursors are impaired. We note, however, that when probes complementary to mature H/ACA snoRNA sequences are used, no obvious accumulation of H/ACA snoRNA precursors was observed. Finally, although the possibility is formally open, impaired production of H/ACA snoRNP proteins is unlikely, for the following reasons: we have established that the steady-state levels of GAR1
mRNAs are not decreased by the time the steady-state accumulation of the proteins that they encode is already significantly diminished. Because Naf1p accumulates within the nucleus, the possibility that Naf1p intervenes in cytoplasmic translation is very slight. A role for Naf1p in the nuclear import of Cbf5p, Gar1p, and Nop10p can still be envisaged.
We favor the idea that Naf1p is involved in H/ACA snoRNP assembly and/or in H/ACA snoRNP transport from the nucleoplasm to the nucleolus. A role for Naf1p in intranuclear trafficking is consistent with its localization in the nucleoplasm and within the nucleolus and with its weak association with mature H/ACA snoRNAs. Intriguingly, the localization of Naf1p in the nucleolus in close vicinity to the dense fibrillar component is reminiscent of that of the nucleolar body, the yeast equivalent of the Cajal body, through which C/D snoRNPs are proposed to transit before reaching the dense fibrillar component (97
). No evidence has been provided so far that H/ACA snoRNPs transit through the nucleolar body. If they do, their routing process to and from the nucleolar body is likely to require trans
-acting factors different from those for C/D snoRNPs (98
). Naf1p is thus a possible candidate.
The presence of some Naf1p within the nucleolus may reflect, rather than a transport function, an involvement of this protein in late H/ACA snoRNP assembly steps. Indeed, in the case of the U3 snoRNP, for example, recent work strongly suggests that final U3 snoRNP assembly occurs in the nucleolar body (in yeast) or the Cajal body (in mammals) (97
). Because Naf1p is predominantly found in the nucleoplasm, a role for this protein in putative early snoRNP assembly steps taking place at or in the vicinity of pre-snoRNA transcription sites is also a very attractive hypothesis. A role for Naf1p in particle assembly is consistent with the observation that depletion of Cbf5p, Nhp2p, or Nop10p, which obviously prevents assembly of complete particles, has the same consequences as does Naf1p depletion on accumulation of H/ACA snoRNAs and Gar1p (29
). In the absence of particle assembly, the body of H/ACA snoRNAs normally protected by bound proteins may be degraded by the exonucleases that remove the flanking sequences and/or may be turned over by a discard pathway distinct from the normal processing pathway. Cbf5p, Gar1p, and Nop10p form a very stable complex (A. Henras and M. Caizergues-Ferrer, unpublished observation) that may be targeted for degradation when stable association with H/ACA snoRNAs is prevented. How could Naf1p promote particle assembly? The most striking motifs present in Naf1p are, in the N-terminal part, a serine-rich domain and, at the C terminus, a large proline- and glutamine-rich domain somewhat reminiscent of a domain found in the RNA-binding protein Nrd1p (80
). Thus, it is conceivable that the C-terminal part of Naf1p could be involved in binding to H/ACA snoRNAs at an early stage. The serine-rich domain of Naf1p, which also contains a high proportion of negatively charged residues, is predicted to be highly phosphorylated by casein kinase II. It could recruit one or several H/ACA snoRNP proteins by providing a binding platform for exposed basic domains of these proteins, such as the C-terminal KKE repeat domain of Cbf5p, the putative amphipathic alpha-helix at the N terminus of Nhp2p, and/or the overall basic small Nop10p protein.
The SMN complex, containing the SMN protein, has also been proposed to be involved in H/ACA snoRNP assembly in some eukaryotes, since the SMN protein can interact with the GAR domains of Gar1p (69
). In addition, the SMN protein has also been shown to interact with the GAR domain of fibrillarin, suggesting that the SMN complex could also be involved in C/D snoRNP assembly (35
). Moreover, a role for the SMN complex, so far not detected in S. cerevisiae
, in the assembly of spliceosomal snRNPs has been well documented in higher eukaryotes (18
). Thus, this complex probably exerts a role that is broader than and different from that of Naf1p. Indeed, the hypothesis that the SMN complex could play in higher eukaryotes the role of Naf1p in lower eukaryotes is contradicted by the fact that Schizosaccharomyces pombe
contains both an SMN complex (27
) and a protein, encoded by the SPBC30D10.15 open reading frame, significantly related to S. cerevisiae
Naf1p (29% identity, 48% similarity over 438 amino acids), which very likely constitutes its S. pombe
orthologue. In addition to SMN, two non-snoRNP proteins, Rvb2p (the yeast orthologue of the mammalian helicase p50) and Srp40p (the yeast orthologue of Nopp140p), have also been proposed to be required, directly or indirectly, for H/ACA snoRNP assembly and localization (36
). Like Naf1p depletion, depletion of Srp40p (in a yeast strain lacking the LES2
gene) inhibits H/ACA snoRNA accumulation, while that of C/D snoRNAs remains unaffected (104
). Unlike Naf1p, however, Srp40p and Rvb2p seem also to be linked to C/D snoRNPs (36
). The mammalian orthologue of Srp40p, Nopp140p, interacts with C/D snoRNP components, and Rvb2p is required for normal accumulation of both C/D and H/ACA snoRNAs. In addition, Rvb2p has been implicated in various processes, including chromatin remodeling (36
). Finally, no interaction has been described between Rvb2p or Srp40p and Naf1p. For all these reasons, Srp40p and Rvb2p are unlikely to directly cooperate with Naf1p. In contrast, a good candidate for a Naf1p partner is the essential Shq1p protein. By a double-hybrid test, Shq1q and Naf1p have been shown to interact (32
). Moreover, Shq1p associates with Flag-Cbf5p (30
) and we have established that Shq1p remains associated (directly or indirectly) with TAP-tagged Gar1p after tandem affinity purification (C. Dez, C. Froment, and V. Dossat, unpublished observation). The functional importance of the interactions detected between Shq1p, Naf1p and H/ACA snoRNP components has in fact been demonstrated by the group of G. Chanfreau, who have shown that Shq1p, like Naf1p, is required for H/ACA snoRNA accumulation (Yang et al., submitted).