To identify binding sites of Prp43 on RNA substrates, we used the CRAC technique, which allows the cloning and sequencing of short RNA fragments that correspond to the binding site(s) of a tagged protein (Granneman et al., 2009
). Crosslinking was either performed in vivo before cell lysis or in vitro after enrichment of Prp43-containing complexes. Since Prp43 has been implicated in pre-mRNA splicing and pre-rRNA processing, we expected to find binding sites in both classes of RNA.
The spliceosomal snRNAs U2, U5, and U6 were previously reported to coprecipitate with Prp43 (Lebaron et al., 2005; Combs et al., 2006; Leeds et al., 2006
), in line with the function of Prp43 in spliceosome disassembly. In our analyses, U6 was significantly crosslinked to Prp43 in vivo, whereas U2 and U5 were not (). This suggests that Prp43 directly contacts U6, while U2 and U5 may associate indirectly with Prp43 as components of a postsplicing complex. No significant enrichment of intronic or exonic sequences was observed for Prp43 relative to the control (, Figure S1
). It may be that interactions between Prp43 and the pre-mRNAs are very transient, or are lost during incubation of the cells on ice prior to and during UV irradiation. Prp43 interacts with the G-patch splicing factor Spp382 (Ntr1) and is a component of the NTR complex (Tsai et al., 2005, 2007; Boon et al., 2006; Tanaka et al., 2007
). This suggests that Prp43 recruitment to the spliceosome might predominately be mediated by protein-protein interactions and U6, with only transient interactions to other RNA species during unwinding.
The large majority of sequences identified for Prp43 in the CRAC approach were mapped to rRNA (). These largely clustered around five putative Prp43-binding sites (), and additional experimental data currently provide strong support for the functional significance of at least three of these sites. The differences in the Prp43 profiles found between in vivo and in vitro crosslinking experiments were unexpected, as previous analyses of the box C/D snoRNP proteins had not revealed similar differences (Granneman et al., 2009
). This presumably reflects functional distinctions between snoRNPs and the helicase. Prior to and during in vivo crosslinking, the living cells are incubated on ice, whereas crosslinking in vitro is preceded by cell lysis, affinity purification on an IgG column, and elution by TEV cleavage, during which Prp43 remains potentially active. We speculate that different interactions are preferentially lost during these different preparations. It is therefore possible that, even though we detect multiple Prp43-binding sites, still further sites have been overlooked in both these analyses. Moreover, snoRNAs that were crosslinked to Prp43 are not all predicted to associate with preribosomes at sites close to the major Prp43 crosslinking sites, making likely that further Prp43-binding sites do indeed exist.
One putative binding site of Prp43 was located at the base of helix 44 in the 18S rRNA sequence, while the others were located in the 25S sequence (). Prp43 was previously shown to coprecipitate with the 20S and 27S pre-rRNAs (Lebaron et al., 2005; Combs et al., 2006; Leeds et al., 2006
), but not the mature rRNAs. This indicates that the helicase is recruited to these regions of the pre-rRNAs prior to final rRNA maturation. Prp43 and its cofactor Pfa1 interact genetically with Ltv1, a component of late pre-40S ribosomes, and with the endonuclease Nob1 that mediates cleavage of the 20S pre-rRNA at site D (Pertschy et al., 2009
). Prp43 was therefore proposed to function in RNA unwinding or structural remodeling of the pre-40S particle, allowing access for the Nob1 endonuclease to site D. The Prp43-binding sites identified by in vivo CRAC at the 3′ end of the 18S rRNA in helix 44 lie close to site D (A), strongly supporting this model.
Four putative binding sites were identified for Prp43 in the 25S sequence; in vivo CRAC revealed one prominent peak (between helices 39/40), whereas in vitro CRAC identified three major Prp43-binding sites (helices 23/24, 34, and 83) (). The major in vivo binding site around 25S+1145 (helices 39/40) is close to modification sites directed by the box H/ACA snoRNA snR5 and the box C/D snR61 (B). There are no data to support an interaction with snR5, but pre-rRNA association of snR61 showed some reduction after Prp43 depletion, and snR61 itself was crosslinked to Prp43 above background levels.
The Prp43-binding site at helices 23/24 lies within the region of 25S that base pairs to the 5.8S rRNA (C). The functional significance of Prp43 binding here has yet to be assessed, but it is conceivable that it plays some role in the establishment of correct 25S-5.8S interactions.
Prp43 bound to 25S rRNA helix 34 appears to function in the release of snoRNAs from pre-60S ribosomal subunits. Six different box C/D snoRNAs, snR39, snR39b, snR50, snR59, snR60, and snR72, with sites of pre-rRNA base pairing close to the Prp43-binding site (see D), were each trapped in association with preribosomes following depletion of Prp43 (). A further box C/D species, snR40, showed a more minor effect, whereas two other snoRNAs that also bind within this general region, snR58 and snR80, were unaffected by Prp43 depletion. This indicates that snoRNA retention does not simply reflect some global defect in preribosome structure. More detailed analysis of this region revealed that rRNA sequences corresponding to the base pairing sites of several snoRNAs of this cluster were enriched with Prp43. These correspond to some of the smaller peaks visible in , such as nucleotides 25S+792–810, which overlap with the base pairing sites of snR39/snR59 (Am807) and snR39b (Gm805), and nucleotides 25S+906–927 for snR60 (Gm908). Moreover, snR72, snR60, and, to a lesser extent, snR40 were directly crosslinked to Prp43. Why this particular subset of the snoRNAs in the cluster was crosslinked is currently unclear. The modification guide sequences of the snoRNAs were specifically enriched for Prp43 crosslinks, and nucleotide substitutions, which indicate the precise crosslinking site, were frequently found in or close to the guide sequences (). These results support the direct unwinding of these pre-rRNA/snoRNA interactions by Prp43.
We propose that the ATPase activity of Prp43 bound to helix 34 in the 27S pre-rRNA specifically mediates the unwinding from the rRNA of several box C/D snoRNAs that have binding sites located in close proximity within the preribosome.
The Prp43-binding site at helix 83 is also located close to a prominent cluster of snoRNA-directed modification sites. The dissociation of snoRNAs in this region was not clearly dependent on Prp43 activity. However, following Prp43 depletion, we observed reduced preribosome association for snR67, which modifies two sites flanking helix 83 (G2619 and U2724). A direct role for Prp43 in snoRNA interactions in the helix 83 region was shown by the identification of numerous crosslinks with the guide region of snR51, which directs modification close to helix 83 at U2729, overlapping an snR67-binding site. In an attempt to better understand these interactions, the profile of crosslinked rRNA sequences obtained for this region was analyzed in closer detail. Only few hits were found from the region immediately encompassing the modified nucleotides 2724 (snR67) and 2729 (snR51), whereas rRNA sequences from the other base pairing site of snR67 around G2619 were strongly enriched with Prp43 (nucleotides 2605–2625). This suggested that Prp43 bound to helix 83 allows access of snR67 to its base pairing site around G2619 in helix 80. However, Prp43 may be redundant with other factors for the access and release of snR51 and snR67 in helix 86.
Prp43 depletion was also associated with reduced preribosome association of snR64, and consistent with this, methylation of C2337 in 25S rRNA was previously reported to be defective in strains lacking Prp43 (Leeds et al., 2006
Even within clusters in which snoRNA binding was affected, depletion or mutation of Prp43 did not affect the binding or release of all snoRNAs predicted to associate with the pre-rRNA in the vicinity of its binding sites. This may reflect a degree of functional redundancy between some of the 19 helicases implicated in ribosome synthesis. Alternatively, different combinations of snoRNAs and helicases may function together at distinct times during ribosome maturation. Within snoRNA clusters, overlapping, mutually exclusive sites of snoRNA binding on the pre-rRNA are common, and there is some evidence for a temporal order in snoRNA binding (Kos and Tollervey, 2005
). It is also possible that at least some snoRNAs are able to bind and dissociate from the pre-rRNA without the aid of helicase activity.
DEAD/H box RNA helicases have been proposed to act nonprocessively in RNA strand displacement, within a radius of their binding site (see, for example, Yang et al., 2007
). The apparent role of Prp43 in promoting the association and/or dissociation of multiple snoRNAs within an rRNA domain would strongly support such a model. To our knowledge, Prp43 is the first RNA helicase involved in eukaryotic ribosome biogenesis for which binding sites on the pre-rRNA have been identified—and a surprisingly complex picture has emerged. Prp43 makes multiple, probably transient and functionally distinct interactions with the preribosomes. Roles in pre-rRNA cleavage and snoRNA binding and release can be assigned to Prp43 proteins bound at distinct individual sites. Whether such striking complexity is common among RNA helicases and/or the many other known ribosome synthesis factors remains to be determined.