Nob1 binds the single-stranded cleavage site D at the 3′-end of 18S rRNA with its PIN domain

Abstract

Ribosome assembly is a hierarchical process that involves pre-rRNA folding, modification, and cleavage and assembly of ribosomal proteins. In eukaryotes, this process requires a macromolecular complex comprising over 200 proteins and RNAs. Whereas the rRNA modification machinery is well-characterized, rRNA cleavage to release mature rRNAs is poorly understood, and in yeast, only 2 of 8 endonucleases have been identified. The essential and conserved ribosome assembly factor Nob1 has been suggested to be the endonuclease responsible for generating the mature 3′-end of 18S rRNA by cleaving at site D. Here we provide evidence that recombinant Nob1 forms a tetramer that binds directly to pre-rRNA analogs containing cleavage site D. Analysis of Nob1's affinity to a series of RNA truncations, as well as Nob1-dependent protections of pre-rRNA in vitro and in vivo demonstrate that Nob1's binding site centers around the 3′-end of 18S rRNA, where our data also locate Nob1's suggested active site. Thus, Nob1 is poised for cleavage at the 3′-end of 18S rRNA. Together with prior data, these results strongly implicate Nob1 in cleavage at site D. In addition, our data provide evidence that the cleavage site at the 3′-end of 18S rRNA is single-stranded and not part of a duplex as commonly depicted. Using these results, we have built a model for Nob1's interaction with preribosomes.

Ribosomes catalyze protein synthesis in all cells. As such, their function and biogenesis have been extensively studied. Classic work in bacteria has characterized the order in which ribosomal proteins (r-proteins) assemble onto ribosomal RNA (rRNA) (1–3). Furthermore, by studying subdomains of the small subunit, it was shown that ordered assembly originates from rRNA conformational changes induced by binding of a subset of r-proteins (4–9). In vivo studies of yeast r-protein assembly suggest rough conservation of the order of r-protein assembly with ribosomes: If the bacterial homolog is a primary binder, depletion leads to an early ribosome assembly defect, while depletion of tertiary binder homologs often results in later assembly defects (10, 11). Despite these similarities ribosome assembly in vivo differs between prokaryotes and eukaryotes. In yeast, over 200 protein and RNA cofactors are required for ribosome assembly (12). Generally, these factors are conserved in eukaryotes but, with few exceptions, are absent in bacteria (13). These assembly factors orchestrate modification and cleavage of the initial 35S precursor rRNA transcript into the mature 18S, 5.8S, and 25S rRNAs, folding of the rRNA, and binding of ribosomal proteins and 5S rRNA. Whereas rRNA modification is well-studied (14, 15), cleavage of the pre-rRNA to release mature rRNAs is poorly understood. Although the order of processing steps has been elucidated, the identity of 6 of 8 endonucleases remains unknown.

Assembly of the small 40S ribosomal subunit in yeast occurs in 2 phases: An early nucleolar stage, comprising transcription, rRNA modification, and cotranscriptional cleavage at site A₂, and a late cytoplasmic stage, during which cleavage at site D occurs (Fig. 1). 20S rRNA, the product of the early steps, is bound by the majority of small subunit r-proteins (10, 11), thereby forming the 43S preribosome. This precursor also contains late assembly factors responsible for export and the final cytoplasmic cleavage at site D (16, 17).

Fig. 1.

(A) rRNA processing pathway in yeast. Cleavage sites are indicated by arrows (red for endonucleases, blue for exonucleases, green if unknown) labeled with the name of the enzyme if known.

The endonuclease responsible for cleavage at site D remains unknown. Nob1 is the best candidate (18, 19), because it is 1 of only 7 nonribosomal proteins bound to 20S rRNA (16). Nob1 also contains a PIN domain, a motif found in some nucleases. A point mutation in the PIN domain is lethal and leads to modest accumulation of 20S prerRNA and strong depletion of 18S rRNA (19). However, point mutations in the putative ATPase Fap7 and the r-protein Rps14 (20, 21) and depletion of the kinases Rio1 and Rio2 and the GTPase-like protein Tsr1 (22, 23) give the same phenotype. Thus, the genetic evidence is not conclusive, and alternative candidates have been presented (20).

PIN domains have structural similarity to the RNaseH superfamily (24–29). In particular, 4 acidic residues at the active site are absolutely conserved. These residues coordinate a divalent metal ion in at least 1 PIN domain (30), and 3 PIN domains have metal-dependent exo- or endonuclease activity, including the nonsense-mediated decay protein Smg6 and Rrp44, the catalytic subunit of the exosome (24, 27, 31, 32). PIN domains exist as monomers, dimers, or tetramers in solution and crystals (24–29). In dimers, the acidic residues that mark the active site face each other (24, 29), whereas the tetramer has 4 active sites around a channel at the center of the complex (24). Interestingly, this channel is wide enough to accommodate single-stranded but not double-stranded RNA. Additional PIN domain-containing proteins in yeast include Utp24, the suggested endonuclease for cleavage at sites A₁ and A₂ (33).

Here, we show that recombinant Nob1 binds as a tetramer to pre-18S rRNA fragments containing cleavage site D. Using RNA binding experiments and dimethyl sulfate (DMS) probing, we show that Nob1 forms interactions with helix 44 (H44) of 18S rRNA, around cleavage site D, as well as with regions close to cleavage site A₂. Moreover, our data indicate that contacts at cleavage site D are made by the PIN nuclease domain. Thus, Nob1 is poised to cleave at site D. These data strengthen the proposal that Nob1 is the endonuclease required for D site cleavage. Additionally, we present data indicating that at this stage in ribosome assembly, cleavage site D is not part of a duplex as commonly drawn, but instead is single-stranded. Combining these data, we have built a model to localize Nob1 on pre-40S ribosomes.

Results

To study its role in ribosome assembly, recombinant yeast Nob1 (51.7 kDa, 459 aa) was purified from Escherichia coli. Mass spectrometry confirmed that Nob1 is the full-length protein, and ICP-MS analysis demonstrates the presence of 1 Zn²⁺ ion per protein, as expected from the Zn-ribbon motif found in Nob1.

Prior studies of small ribosomal subunit assembly have taken advantage of the domain organization of this large RNA. Individual RNA domains bind proteins and fold into tertiary structures indistinguishably from the same structures embedded in the entire subunit (4–9). This has greatly facilitated in vitro studies of ribosome assembly, as it allowed problems to be attacked in a stepwise fashion. Furthermore, individual folding and assembly steps were successfully isolated and dissected in that manner. Building on this strategy, we obtained pre-18S rRNA fragments that include the 3′-minor domain of 18S rRNA and parts of internal transcribed spacer 1 (ITS1) (Fig. 2A) by in vitro transcription. The ends of these RNAs were chosen because prior work had either suggested that they are located at the ends of secondary structure elements or shown that rRNA pieces ending at these sites were functional in vivo (34, 35). Furthermore, similar fragments form in vivo when cleavage at site D is inhibited (18, 21). We thus rationalized that these fragments would fold into unique structures relevant to in vivo ribosome assembly as confirmed by native gels and DMS probing experiments (see below). DMS probing also shows that the sequence contained in the mature rRNA folds into the secondary structure elements observed in the crystal structures of the mature small ribosomal subunit (36, 37).

Fig. 2.

(A) Pre-rRNA fragments used in this study. 5′- and 3′-ends are labeled in green or red, respectively. Thus, the H44/A₂ RNA starts at the green arrow labeled H44 and ends at the red arrow labeled A₂. The point of truncation for H44Δ is indicated with the dashed line. Residues indicated in blue are mutated for the experiments in Table 2. (B) Nob1 binding to H44/A₂ (●) and H44/18S pre-rRNA (○) fragments. Data were normalized and fit with Eq. 1. Three independent experiments yield values of K_d = 0.24 ± 0.05 and 3.5 ± 1.6 μM⁴ for H44/A₂ and H44/18S, respectively. (C) Stoichiometry of Nob1·pre-rRNA interaction. The intersection point yields the stoichiometry.

Nob1 Binds rRNAs Comprising Cleavage Site D.

Reasoning that any enzyme must be able to bind its substrate, we tested whether Nob1 bound to pre-18S rRNA. Native PAGE demonstrates that Nob1 binding shifts pre-rRNA fragments comprising cleavage site D (H44/A₂) to lower electrophoretic mobility, indicating an interaction (Fig. S1A). This interaction is specific, as only unlabeled H44/A₂ RNA but not polyU RNA competes for this interaction (Fig. S1B). In contrast, the interaction between Nob1 and the thiostreptone loop (TSL), a 25S rRNA fragment similar in size to the pre-18S fragments used herein, is much weaker and competed off by H44/A₂ and polyU rRNA (Fig. S1B), indicating that it is nonspecific. Together, these data provide strong evidence that Nob1 forms direct and specific interactions with pre-18S rRNAs containing cleavage site D. We also tested whether Fap7, the other protein suggested to be the nuclease for D-site cleavage (20), binds D-site-containing RNAs and found no evidence for any interactions up to 20 μM Fap7.

Nob1 Binds at Cleavage Site D and in ITS1.

To delineate Nob1's binding site, we compared binding of a number of rRNAs ending at sites D, A₂, as well as just upstream of site A₃ (Fig. 2 A and B, and Table 1, upper panel). Our data show that Nob1 binding to the substrate analog comprising the 3′-minor domain of 18S rRNA and ending at cleavage site A₂ (H44/A₂) is >10-fold stronger than binding to a product analog (H44/18S), which terminates at the 3′-end of 18S rRNA (Fig. 2B and Table 1), as expected from an enzyme, which should bind substrate but release product. Further lengthening the RNA to end at +228 increases binding another ≈3-fold. A further 2-fold increase is seen from the addition of another hairpin to end at +278. These results show that elements in ITS1 both 5′ and 3′ to site A₂ contribute to Nob1 binding. This finding was at first unexpected, as the in vivo substrate for Nob1 ends at site A₂. However, in vivo, Nob1 binds to an early pre-rRNA precursor that includes this element (18), perhaps via this interaction.

Table 1.

Nob1 binding to pre-18S rRNA fragments

RNA (5′ end / 3′ end)	Kd, μM4
H45/18S (0)	>25*
H45/A2 (+212)	1.8 ± 0.5
H45/(+228)	0.50 ± 0.10
H45/(+278)	0.28 ± 0.05
H45(−31)/A2	1.8 ± 0.5
H44(−164)/A2	0.24 ± 0.05
H44Δ/A2	0.18 ± 0.05

Data are the averages of at least 3 experiments.

*Nob1 appears to bind this RNA as a dimer, so accurate comparisons to other data cannot be made.

Comparison of rRNAs starting at the 5′-end of helix 44 (H44) and at the 5′-end of helix 45 (H45) shows that addition of H44 contributes ≈8-fold to Nob1 binding (Table 1, lower panel). Replacing the lower part of H44 with GAGA (H44Δ/A₂) has no effect on Nob1 binding, suggesting that Nob1 only contacts the upper part.

To confirm these results, we probed the accessibility of rRNA to DMS in the absence and presence of Nob1. DMS methylates the N1 of adenosine and the N3 of cytosine unless they are protected by the formation of base pairs, tertiary structure, or protein binding (38). Methylation is detected by the ensuing reverse transcription stop. By comparing the modification pattern in the absence and presence of protein, one can pinpoint sites protected due to protein binding. These sites of protection can arise from direct binding of protein or from protein-dependent rRNA conformational changes. Because base pairing prevents DMS methylation, these experiments also provide information about rRNA secondary structure.

DMS probing of pre-rRNA fragments alone shows that they fold into unique structures with distinctly protected and accessible sites (labeled in yellow in Fig. 3 and Fig. S2B). The modification pattern in the 3′-minor domain is consistent with its secondary structure in mature ribosomes (36, 37) demonstrating that our rRNA fragments fold stably into the expected secondary structure. Our protection data also suggest that the region between cleavage sites D and A₂ folds into an extended duplex containing several bulges, consistent with prior data (35).

Fig. 3.

DMS protection data mapped onto the secondary structure of H44/A₂ pre-rRNA. Sites indicated with yellow dots and circles are strongly and weakly modified by DMS, respectively. Sites indicated by red-filled yellow circles are modified by DMS in the absence but not the presence of Nob1. Sites shown in gray give strong reverse transcription stops in the absence of DMS and thus cannot be interpreted. Examples of protection data in the ITS1 loop (top), around cleavage site D (middle), or within H44 (bottom).

Nob1 binding to rRNA protects a number of sites that cluster around cleavage site D, as expected from an endonuclease cleaving at site D (red-filled yellow circles in Fig. 3). In addition, protections are seen at the top of H44, in a conserved loop between cleavage sites D and A₂, and adjacent to site A₂. No protected site becomes accessible, which would indicate conformational changes. Importantly, these protection data are fully consistent with our biochemical data, which indicate interactions in H44 as well as beyond cleavage site D. We cannot rule out additional interactions with Nob1 in the double-stranded regions of ITS1, as these residues are protected in the absence of Nob1 and thus silent in our DMS protection assay.

To test whether Nob1 also protected these sites in vivo, we generated a galactose-inducible Nob1 strain. This strain contains Nob1 when grown in the presence of galactose, but Nob1 is depleted ≈8 h after shift to glucose, similar to published results (18). Importantly, we found that the same residues around the cleavage site protected by Nob1 in vitro are protected in the presence of Nob1 in vivo (Fig. S3). We were unable to probe residues in H44 because of the strong reverse transcription stop at methylated adenosines in H45. Residues close to site A₂ could not be probed, as the required primer anneals only in the 35S precursor and gives poor signal, likely because 35S rRNA is less abundant and more heterogeneous. While we cannot rule out indirect effects, these seem unlikely given the observation that depletion of Nob1 leads to accumulation of 20S rRNA without affecting other processing steps (18). If depletion of Nob1 resulted in loss of additional assembly factors, such as Dim1, Dim2, Enp1, or Ltv1 (and these were binding at the cleavage site), loss of Nob1 would be expected to mirror the early assembly defect observed upon deletion of these factors. Thus, the simplest interpretation is that Nob1 binds at cleavage site D in vivo.

Nob1's PIN Domain Binds at Cleavage Site D.

Nob1-dependent protections around cleavage site D could arise from interactions with its PIN domain or other regions of the protein. Yet, cleavage is only expected to occur if the PIN domain, containing the active site, binds at site D. To test whether protections around site D arise from the PIN domain, we cloned and purified Nob1 fragments containing just the PIN domain [Nob1 (1–159) and Nob1 (1–194); PIN domain homology ends ≈amino acid 130] and confirmed that these retain RNA binding affinity (Fig. S4A). The PIN domains protect a subset of the residues protected by full-length Nob1, including all of the residues at the cleavage site (shown in red in Fig. S4D). Residues in H44 protected by full-length Nob1 are also protected, along with additional residues in H45 (shown in pink). These protections may arise from the MBP-tag that these fragments retain to keep them soluble or from the 30–60 residues outside the PIN domain. These results are consistent with the PIN domain contacting the cleavage site.

Nob1 Binds rRNA as a Tetramer.

The data in Fig. 2B show that Nob1 binding to rRNA is cooperative. This was surprising as gel-filtration indicated that Nob1 is a monomer. Prior results showed that PIN domains could form monomers, dimers, or tetramers (24–28). To test whether Nob1 formed multimers in the presence of RNA, we determined the stoichiometry of the Nob1·RNA complex. To ensure that the Nob1·RNA complex was saturated by RNA regardless of the amount of protein added (so that each protein molecule binds an available RNA molecule), trace amounts of labeled RNA were added to unlabeled RNA at concentrations at least 4- to 5-fold above the K_d value. Protein concentrations were varied from sub- to superstoichiometric. As shown in Fig. 2C and Fig. S1C, a complete shift of RNA from the free to the bound form requires 4 equivalents of protein, providing strong evidence for a 4:1 ratio of Nob1 to RNA. The simplest interpretation of this result is that Nob1 forms a tetramer in the presence of RNA, although an 8:2 complex cannot be ruled out. This stoichiometry did not change for any of the 3 RNAs tested (H44/A₂, H44/+245, H45/+245)*. Formation of a Nob1 tetramer is also consistent with the observed average Hill coefficient of 3.2, which provides a lower limit for the stoichiometry of the interaction^†.

Cleavage Site D Is Single-Stranded.

The structure of the tetrameric PIN domain from Pyrobaculum aerophilum PAE2754 shows 2 dimers sandwiched onto each other, rotated by 90 °, forming a structure in which the active sites of each molecule, indicated by conserved acidic residues, face the interior of the complex, accessible only from a small channel along the center of the cavity (figure 2 in ref. 24). The dimensions of the channel are large enough to accommodate single-stranded but not double-stranded RNA. As described above, our data suggest that Nob1 is tetrameric when bound to RNA. If tetrameric Nob1 bound to rRNA forms a similar structure to other PIN domains (modeled in ref. 19), then cleavage site D must either be single-stranded in 20S rRNA or, alternatively, the duplex commonly believed to form must be dissociated by binding of Nob1. We used RNA binding experiments with RNA mutants in which a possible duplex is disrupted or strengthened and DMS probing to distinguish between these possibilities.

If Nob1 binding requires duplex dissociation then the observed energy of Nob1 binding would be the sum of the energy from Nob1 interacting with RNA and the energy required to dissociate the duplex. Based on nearest-neighbor calculations, changing 4 A·U pairs into G·C pairs (blue residues in Fig. 2A) stabilizes a duplex 50,000-fold (39). Thus, if Nob1 binding was coupled to duplex dissociation, Nob1 binding is expected to be ≈50,000-fold weaker for a G·C duplex. Conversely, mutations on 1 side of the duplex alone should weaken or disrupt a duplex, thereby strengthening Nob1 binding. Mutations 5′ to the D-site weaken binding of Nob1 ≈7-fold (“mut D” in Table 2), and mutations 3′ to the D-site weaken binding 3-fold (“mut αD” in Table 2). These data indicate that Nob1 makes functionally important contacts with residues adjacent to the cleavage site, consistent with the observed DMS footprint at that site (Fig. 3). Combining both mutations to stabilize a duplex is additive, but not more than additive, and much less than expected if a 50,000-fold stronger duplex has to be disrupted for Nob1 binding. These data are inconsistent with Nob1 disrupting a duplex before binding the cleavage site. Thus, either the proposed binding of the cleavage site at the center of a PIN tetramer is incorrect, or the cleavage site is single-stranded even in the absence of Nob1.

Table 2.

Nob1 binding to D-site mutants

	wild type	mut D	mut αD	Both
Kd, μM4	0.35 ± 0.05	2.3 ± 0.2	1.1 ± 0.1	6.7 ± 0.4
Krel	(1)	6.6	3.1	19.1

Data are averages of at least 3 experiments. mutD: (−3)UUAAA→CUGCG; αmutD: (7)UUUAA→CGCAG (Figure 2A).

To further test whether the cleavage site was single- or double-stranded, we reinspected the DMS probing results. These experiments show that in the absence of Nob1, every adenosine, as well as the cytosine around cleavage site D, are strongly methylated by DMS (Fig. 3). These results provide strong evidence that the rRNA around cleavage site D is single-stranded in our system.

To test whether the cleavage site was also single-stranded in the 20S rRNA precursor that accumulates in vivo, we carried out in vivo footprinting of the sequence around the cleavage site. Fig. S3 shows that the RNA around site D is accessible to DMS in the absence of Nob1, suggesting strongly that it is single-stranded in vivo.

Discussion

Cleavage Site D Is Single-Stranded in 20S rRNA.

The PIN-domain exonuclease PAE2754 forms a tetramer (24). When the PIN domain from Nob1 is modeled onto this structure (Fig. 1 in ref. 19) the C-terminal ends point outward, suggesting that this model could represent the structure of the PIN domains in the tetrameric full-length Nob1 we observe in biochemical experiments. Because in this model the active site is found at the interface of the subunits around a channel wide enough to accommodate single-stranded but not double-stranded RNA, our results that Nob1 is tetrameric combined with the previous modeling suggests that the RNA around cleavage site D must be single-stranded to be cleaved by Nob1. Because common models for the RNA structure around cleavage site D show a duplex (as shown in Fig. 2A), we first tested whether Nob1 unwound a duplex. Analysis of RNA binding to mutant RNAs in which a putative duplex is either strengthened or weakened provides strong evidence against a model in which Nob1 binding is coupled to duplex unwinding. Furthermore, DMS protection data show directly that the RNA around cleavage site D is single-stranded in vivo and in vitro. This finding is consistent with the observation that the ability to form a duplex at site D is not conserved in eukaryotes (40, 41). Furthermore, mutations that would disrupt a duplex have no effect on rRNA maturation or stability (42).

A duplex around cleavage site D in eukaryotic preribosomes would be located in the same place as the Shine Dalgarno (SD) duplex in prokaryotes. Formation of the SD duplex fixes the conformation of the head and body with respect to each other and prealigns ribosomes for interaction with tRNA^fMet (36, 43). Given the conservation of ribosomes, formation of a duplex around cleavage site D, positioned as the SD duplex, could lead to similar conformational rearrangements, thereby promoting premature binding of initiator tRNA to eukaryotic preribosomes. Thus, the single-stranded D-cleavage site may have evolved to prevent premature translation initiation, akin to KsgA (Dim1 in yeast), whose binding prevents IF3 binding to ribosomes (44).

Our conclusion that cleavage site D is single-stranded differs from models that show cleavage site D as part of a duplex (34, 42, 45). Inspection of prior data indicates that these models are based more on secondary structure predictions and less on the data in the original publications, which show the region in question to be accessible to DMS (35, 46). Furthermore, Yeh et al. show only uridines 7–10 in ITS1, but not the “opposing” adenosines, to be cleaved by RNase V1, inconsistent with duplex formation between these 2 regions (35). However, an alterative structure in the 35S precursor may allow for formation of a duplex early in ribosome assembly. This duplex would then have to be unwound before or soon after formation of 20S rRNA to allow for Nob1 binding and cleavage at site D.

The Nob1 Binding Site in Preribosomes.

Nob1 is part of a well-defined pre-40S ribosomal complex that contains the 20S precursor to 18S rRNA and most small subunit r-proteins (10, 11). In addition, this precursor contains 6 ribosome assembly factors: The methylase Dim1, the RNA binding protein Dim2, the kinase Rio2, the export factor Ltv1, as well as Enp1, and Tsr1 (16). The binding site for bacterial Dim1 on the mature small subunit has recently been located (44). Here, we have mapped the Nob1 binding site on pre-18S analogs. DMS footprinting shows that the Nob1 binding site is centered around cleavage site D at the 3′-end of 18S rRNA. In addition, we have shown that Nob1 interacts with the top of H44 as well as within ITS1. These sites of interaction are highlighted in spacefill in Fig. 4 (only the first 7 residues in ITS1 are modeled). Our results also suggest that the protections around the cleavage site arise from the PIN domain. Furthermore, we have shown that Nob1 forms a tetramer, and analysis of RNA binding of the PIN domain alone suggests that this domain is responsible for tetramer formation, consistent with prior results showing that PIN domains can form multimers (24, 29). Using these results and the finding that the cleavage site is single-stranded, we have built a model for the interaction of Nob1's PIN domain with preribosomal particles (Fig. 4). While this model is not believed to accurately predict the molecular details of this interaction, it suggests interactions between Nob1 and the head and platform region of the small subunit, formed by the 3′-major and central domain, respectively. We have tested this prediction and find that adding the 3′-major domain contributes 6-fold and addition of the central domain another 4-fold to RNA binding (Table S1), confirming the model. In addition, we have tested for interactions between Nob1 and Rps14 (S11 in bacteria, red), Rps0 (S2 in bacteria, pink), and Rps5 (S7 in bacteria, salmon) using pulldown experiments. All of these proteins are located in the vicinity of the PIN domain. Fig. S5 shows that Nob1 binds MBP-tagged Rps5 and Rps14 but not MBP-tagged Rps0, providing further evidence for the suggested location of Nob1 on the preribosome.

Fig. 4.

Model for the interaction of Nob1's PIN domains with preribosomes. Sites protected by binding of Nob1 are shown in space fill, rRNA is in gray, H44 is labeled in cyan, H45 is shown in blue, and the remainder of the 3′-end in dark blue. The 2 nucleotides on either side of the cleavage site are shown in green. Nob1's PIN domains are shown in 4 shades of yellow. Ribosomal proteins are shown in purple, except for S11 (Rps14, red), S2 (Rps0, pink), and S7 (Rps5, salmon). The inset shows the entire subunit from the view of the large subunit.

The protections in H44 are not accounted for by the model in Fig. 4, even though we observe protections in that region from the Nob1-PIN fragments. These protections may arise from the MBP-tag contained in those fragments (but not full-length Nob1), consistent with the protections in H45 (Fig. S4D). Moreover, ≈30–60 aa contained in the Nob1-PIN fragments are not modeled due to lack of homology. Our data indicate that these regions interact with RNA and are required for site-specific binding (Fig. S4A), consistent with the large number of positively charged residues in that part of the protein. These residues may also account for the protections in H44. Alternatively, it is possible that positioning of Nob1 requires additional movements in the small subunit that bring the 3′-end closer to H44. Finally, protections in H44 could result from its Zn-ribbon domain.

How Is Nob1 Activated?

We have shown that Nob1 binds to cleavage site D in vitro and in vivo. Moreover, our data indicate that these protections arise from binding of the PIN domain, the proposed active site of the protein, based on the observed nuclease activity of other PIN domain-containing enzymes (24, 27, 31, 32), the observation that a mutation in a conserved acidic residue abolishes cleavage at site D in vivo (19), and the ability of Nob1's PIN domain to substitute for the PIN domain of Smg6 in nonsense-mediated decay (31). Together these data provide strong biochemical data linking Nob1 to cleavage site D, where it is poised to carry out cleavage. Nevertheless, we have not yet been able to observe Nob1-dependent rRNA cleavage. Only 3 of the 7 tested PIN domains have activity by themselves in vitro (24, 27, 31, 32), suggesting that PIN activity may be modulated by accessory factors. Such factors in ribosome assembly include Dim2, a suggested binding partner of Nob1 (47). Furthermore, the C-terminal tail of S11 (Rps14) winds around to the 3′-end of 16S rRNA in the crystal structure (Fig. 4), and a point mutation in the tail of Rps14 abolishes cleavage at site D in yeast (21).

Materials and Methods

RNA Binding.

rRNA was folded in the presence of 10 mM Mg²⁺ as described (48). Trace amounts of radiolabeled rRNA (<16 nM) were incubated with varying concentrations of Nob1 for 2 h in 100 mM KCl, 50 mM HEPES, pH 7.5, 10 mM MgCl₂ at 30 °C and loaded onto a 6% acrylamide/THEM (33 mM Tris, 67 mM HEPES, 1 mM EDTA, 10 mM MgCl₂) gel (49). Protein-bound and free fractions were quantitated using phosphoimager analysis, and data were fit to Eq. 1 using Kaleidagraph (Synergy Software). This yields the overall K_d as described in the SI Methods. No differences were observed between 4 Nob1 preparations.

To determine the stoichiometry of the Nob1-rRNA interaction, 4 μM of rRNA (H44/A₂, H44/+245, or H45/+245) spiked with trace amounts of the same radiolabeled rRNA were incubated with increasing concentrations of Nob1 and separated by native PAGE as described above.

DMS Probing.

DMS probing was performed as described (50). Unlabeled H44/A₂ or H45/+278 rRNA (2 μM) was folded and incubated in the presence or absence of 10 μM Nob1, exposed to 2.3% DMS for 3 min, before quenching with β-mercaptoethanol. Phenol-extracted RNA was annealed to radiolabeled primers u–z for reverse transcription. Sequencing lanes were obtained by dideoxy sequencing of rDNA using Sequenase 2.0 (USB). The protections and modifications shown in Fig. 3 are reproducible in at least 3 experiments, conducted with 2 different RNA samples. To account for small differences in the amount of radioactivity in each lane, we “normalized” to the neighboring bands in the same lane. In ambiguous cases, a line-scan was performed using ImageQuant.

Structure Modeling.

The 30S subunit from Thermus thermophilus [(36), 2E5L] was used for modeling. mRNA, S6, and S18 were removed (these proteins have no eukaryotic homologs). T. thermophilus has 16 nucleotides 3′ to H45, whereas yeast 18S rRNA contains 9. Thus, we labeled the bond located 9 residues 3′ to H45 as the scissile bond, which places the remaining residues in ITS1. Using Pymol, these residues were shifted into an extended conformation, reasonable for single-stranded RNA. The model for a tetramer of Nob1's PIN domains (19) was manually docked this structure to minimize clashes, account for protections, and place the putative active site in proximity of the D-cleavage site.

Additional methods are described in SI Methods.