The yeast ATPase Prp43p has been shown to be required, at least in vitro, for the last step of splicing, that is, the dissociation of the late postsplicing complex from the spliced-out intron in the lariat form (1
). Consistent with this, we find that the U2, U5, and U6 spliceosomal snRNAs are precipitated with yeast Prp43p. Interestingly, we find that yeast Prp43p interacts with several components of the Prp19p-associated complex (nineteen complex or NTC) that is required for spliceosome activation (8
), suggesting that yeast Prp43p may have additional, earlier roles in splicing. This seems at any rate to be the case in humans, since the likely human orthologue of yeast Prp43p, hPrp43, was detected in affinity-purified pre-spliceosome complex A (39
), spliceosome complex B prior to activation (BΔU1) (60
), activated B* spliceosome poised for splicing (59
), and spliceosome complex C (48
) that has undergone the first trans
-esterification reaction. Yeast Prp43p may even associate cotranscriptionally with nascent pre-mRNAs, because tandem affinity purification of Prp43p-TAP associated with mass spectrometry analysis revealed an interaction of this protein with a surprising large number of chromatin remodeling-RNA polymerase II-associated factors (Table ). In addition to its involvement in splicing, we propose that Prp43p also plays major, direct roles in eukaryotic ribosome biogenesis. We show in the present report that Prp43p depletion leads to a dramatic drop in the steady-state levels of all pre-rRNAs, except the 35S pre-rRNA, resulting in greatly diminished levels of all mature rRNAs.
While we cannot formally exclude the possibility that this inhibition of ribosome biogenesis is solely the indirect consequence of splicing defects, we deem it unlikely for the following reasons. In the case of both yeast and human cells, Prp43p is detected in the nucleolus (26
), the nuclear locale devoted primarily to early steps of ribosome biogenesis. We demonstrated by a very thorough study combining immunoprecipitations, TAP coupled to mass spectrometry, and double-hybrid analysis that yeast Prp43p is a component of almost all preribosomal particles, namely, 90S, pre-40S, and pre-60S particles. Our results extend those obtained during earlier affinity purifications (33
) and a double-hybrid study (46
) that found Prp43p associated with a few components of 90S and pre-60S preribosomal particles. The highly significant link between Prp43p and some components of these particles has also been underscored by a recent bioinformatic analysis (55
). Crucially, we note that under the experimental conditions we used, the depletion of Prp43p has little effect on the steady-state accumulation of some mature mRNAs that are produced by splicing, such as the mRNAs encoding actin or cytoplasmic light chain dynein. Thus, we think it rather unlikely that a defect in the splicing process per se did cause such a substantial drop in the steady-state levels of mRNAs for ribosome biogenesis factors, and hence a significant reduction of their de novo synthesis, that it would indirectly have led to the major perturbation of ribosome synthesis we observe. We did check the steady-state levels of ribosomal protein mRNAs and found that they declined sharply during Prp43p depletion. However, this drop is not due to splicing inhibition because levels of ribosomal protein mRNAs that are produced by splicing and of those that are not were diminished equally. This phenomenon may reflect a feedback mechanism, transcriptional and/or posttranscriptional, that shuts down ribosomal protein expression when synthesis of both small and large ribosomal subunits is severely compromised.
Prp43p is one of the few known ribosome biogenesis factors required for the synthesis of both small and large ribosomal subunits and present within pre-40S and pre-60S particles. Other examples include Rrp5p, required for the synthesis of 18S rRNA and the short form of 5.8S rRNA (22
), Rrp12p, which intervenes in the export of both small and large preribosomal subunits from nucleus to cytoplasm (66
), and the DEAD-box ATPase Has1p, which is needed for 18S rRNA synthesis but is also found within pre-60S preribosomal particles (21
). Prp43p is also one of the rare examples of factors implicated in both splicing and ribosome biogenesis and is to our knowledge the only helicase determined as such so far. The most thoroughly studied factor shared between the splicing and ribosome biogenesis processes is Snu13p, a protein component of U4 snRNP and of all box C/D snoRNPs (97
). Whether the use by the cell of Prp43p in both splicing and ribosome biogenesis contributes to a hypothetical coordination between these two processes remains to be determined.
The finding that Prp43p is associated with components of the RNA polymerase I machinery suggests that Prp43p assembles with nascent 90S preribosomal particles during pre-rRNA transcription. However, contrary to the findings with respect to the tUTP components of the U3 processome that are required for RNA polymerase I transcription (20
), Prp43p does not play a fundamental role in that process because the 35S pre-rRNA accumulates in Prp43p-depleted cells while it disappears in tUTP-depleted ones. We also observe a clear association of Prp43p with cytoplasmic 18S rRNA. The precipitation efficiency of 25S rRNA with tagged Prp43p is low but is nevertheless clearly above background levels. We also detect an association between Prp43p and 5.8S and 5S rRNAs when the protein is overexpressed. Moreover, Prp43p-TAP interacts with Lsg1p, a cytoplasmic GTPase required for a remodeling step of 60S ribosomal particles in the cytoplasm (40
), arguing that Prp43p is associated with mature or nearly mature 60S ribosomal particles in the cytoplasm. Thus, it seems that Prp43p remains part of the ribosome biogenesis process from start (pre-rRNA transcription) to finish (production of mature 40S and 60S ribosomal particles in the cytoplasm). The prolonged association of Prp43p with preribosomal particles may seem counterintuitive for an enzyme proposed to drive transient conformational rearrangements. It is, however, a phenomenon already well documented in the case of the potential helicases Sub2p (32
) and Dbp5p (24
), which are probably recruited to nascent pre-mRNA during transcription and remain linked to the transcript throughout splicing. Strikingly, in similarity to what we find in the case of Prp43p and preribosomal particles, Dbp5p accompanies mRNAs during export through the nuclear pore into the cytoplasm, where it drives mRNP rearrangements. The prolonged presence of Prp43p within preribosomal particles may reflect a structural role for this protein. It is also possible that Prp43p drives a succession of conformational rearrangements through several rounds of ATP hydrolysis and several association-dissociation cycles. In that respect, it is interesting that the dominant-negative forms of Prp43p, although predominantly stalled in 90S preribosomal particles, are also found in more downstream particles, suggesting that Prp43p may “enter” at several points along the pathway. The precise functions and molecular substrates of Prp43p remain to be identified. Our data suggest that the ATPase activity of this protein is required early, already within 90S preribosomal particles, to allow efficient processing of the 35S pre-rRNA. Since Prp43p associates with both pre-40S and pre-60S preribosomal particles, it must have several partners and substrates. Within pre-40S and mature or nearly mature 40S particles, and only there, Prp43p may be more closely linked to the Pfa1p protein. This link is, however, not a crucial one, because the growth disadvantage caused by lack of Pfa1p can only be detected under mixed-culture conditions. Clearly, identifying the direct partners and substrates of Prp43p constitutes a challenge for future research.