|Home | About | Journals | Submit | Contact Us | Français|
How pseudouridylation (Ψ), the most common and evolutionarily conserved modification of rRNA, regulates ribosome activity is poorly understood. Medically, Ψ is important because the rRNA Ψ synthase, DKC1, is mutated in X-linked Dyskeratosis Congenita (X-DC) and Hoyeraal-Hreidarsson syndrome (HH). Here we characterize ribosomes isolated from a yeast strain where Cbf5p, the yeast homologue of DKC1, is catalytically impaired through a D95A mutation (cbf5-D95A). Ribosomes from cbf5-D95A cells display decreased affinities for tRNA binding to the A- and P-sites as well as the cricket paralysis virus IRES (Internal Ribosome Entry Site), which interacts with both the P- and E-sites of the ribosome. This biochemical impairment in ribosome activity manifests as decreased translational fidelity and IRES-dependent translational initiation, which are also evident in mouse and human cells deficient for DKC1 activity. These findings uncover specific roles for Ψ modification in ribosome-ligand interactions that are conserved in yeast, mouse, and humans.
Pseudouridine (Ψ), a C-glycoside isomer of uridine, is the most common single nucleotide modification found in functional RNA and often appears in highly conserved regions of homologous RNAs (Ofengand, 2002). The S. cerevisiae ribosomal RNA (rRNA) contains 44 Ψ residues (Liang et al., 2009). Several lines of evidence suggest an important role for base modification in ribosome activity: Ψ formation is targeted only to residues that compose the mature ribosome (Jeanteur et al., 1968); Ψ bases are conserved from bacteria to mammals and the number of Ψ bases in the ribosome increases dramatically from eubacteria to eukaryotes (Ofengand, 2002); and Ψ residues tend to cluster in functionally important regions of the ribosome (Decatur and Fournier, 2002). It has been hypothesized that Ψ is an essential component of proper RNA folding and function (Ofengand, 2002). However, when Ψ residues are removed individually or in groups, often little to no effect is seen on growth rates (Charette and Gray, 2000). Still, Ψ mutants are often strongly outcompeted by their wild-type counterparts, indicating an advantage incurred by their presence (Baxter-Roshek et al., 2007).
Cbf5p, the yeast pseudouridine synthase, and dyskerin (DKC1), its mammalian homolog, are the catalytic core of the larger H/ACA ribonucleoprotein complex (H/ACA RNP), named for the family of structurally conserved small nucleolar RNAs (snoRNAs) that serve to guide sequence specific rRNA pseudouridylation (Liang and Li, 2011). H/ACA RNPs contain three additional highly conserved proteins: Nop10p, L7Ae/Nhp2p, and Gar1p. The overall structure of the H/ACA RNP is “T-shaped”, where the archaeosine transglycosylase (PUA) domain of Cbf5p/Dyskerin is the left arm of the T, the catalytic core of the protein is in the center, Nop10p and L7Ae are the right arm, and Gar1p anchors the bottom. The Box H/ACA snoRNA runs horizontally across the center of the structure participating in multiple interactions with the PUA domain and the catalytic core of Cbf5p. Nop10p has an elongated structure that tightly binds Cbf5p, stabilizing its active site structure (Hamma et al., 2005). L7Ae binds Nop10p and specifically interacts with the K-turn and K-loop structural motifs of the Box H/ACA snoRNA. Furthermore, a considerable degree of structural rearrangements occur during the sequential assembly of the RNP (Hamma and Ferre-D’Amare, 2010). In mammals, dyskerin has additional functions as a component of the telomerase complex, and a possible role in mRNA splicing through direct association with small Cajal body RNAs (scaRNAs) (Meier, 2005). Mutations in the DKC1 gene have been linked to X-linked Dyskeratosis Congenita (X-DC), a disease characterized by bone marrow failure, skin abnormalities and increased cancer susceptibility (Alter et al., 2009; Dokal, 2000). Most DKC1 mutations lie in the PUA domain (Liang and Li, 2011; Hamma and Ferre-D’Amare, 2010), but two mutations found in the catalytic domain are associated with a severe variant of X-DC known as Hoyeraal-Hreidarsson (HH) syndrome (Dokal, 2000). Hypomorphic DKC1 mutant mice (Dkc1m) show decreased Ψ content in their rRNAs, and impairments in Ψ affect translational control of specific mRNAs harboring IRES elements (Yoon et al., 2006; Ruggero et al., 2003). Defects in translational control have also been observed in X-DC human patient cells (Yoon et al., 2006; Bellodi et al., 2010a; Bellodi et al., 2010b). The molecular mechanisms by which ribosomes lacking Ψ modifications are impaired in translational control remain poorly understood.
While CBF5 is an essential gene, high copy expression of a cbf5 allele in which the conserved aspartate residue in the catalytic domain required for pseudouridylation was changed to alanine (cbf5-D95A) resulted in viable yeast that lack rRNA Ψmodification, revealing that these are dispensable for cell survival (Zebarjadian et al., 1999). This is in contrast to the only essential yeast H/ACA RNA, snR30, which does not direct Ψ. Thus, expression of the cbf5-D95A allele should not affect assembly of H/ACA RNPs with snR30, raising important questions regarding the origin of the rRNA processing defects that could contribute to a growth defect of these cells. In the current study, we use yeast containing this same mutant cbf5-D95A expressed from a low-copy episomal plasmid in cbf5Δ cells, cells isolated from Dkc1m mice, and siRNA knockdown of DKC1 in human cells to examine the effects of Ψ rRNA defects on ribosome biochemistry and translational fidelity. Together, this analysis reveals the importance of rRNA pseudouridylation for ribosome ligand binding and translational fidelity that are evolutionarily conserved from yeast to human cells.
Translation directed by the CrPV IGR IRES element is impaired in Dkc1m hypomorphic mice and X-DC patient cells, and deregulation of CrPV IGR IRES-mediated translation is associated with significant reductions in total rRNA Ψ levels (Yoon et al., 2006). To further study the molecular mechanisms underpinning this defect, we employed the yeast cbf5-D95A mutant (hereafter referred to as D95A), which lacks all Ψ in the rRNA. Translation initiation from the CrPV IGR IRES was monitored using a dual-luciferase reporter which is able to mediate cap-independent translation in yeast cells (Thompson et al., 2001) (Figure 1A). While overall cap-dependent translation was not affected by the D95A mutation, CrPV IGR IRES-mediated translation was reduced by more than 90% in the D95A as compared to wild-type (WT) cells (Figure 1B).
The CrPV IGR IRES can directly bind the 40S ribosomal subunit (Jan and Sarnow, 2002; Spahn et al., 2004; Pestova et al., 2004; Landry et al., 2009), and in vitro and in vivo assays suggest that the CrPV IGR IRES can initiate translation in the absence of any initiation factors (Jan and Sarnow, 2002; Pestova et al., 2004). It has been speculated that translation initiation can be partially explained by the fact that the CrPV IGR IRES forms a highly complex tertiary structure and one domain of this IRES element assumes a tRNA-like conformation that interacts with the ribosomal P-site (Costantino et al., 2008). Steady-state filter-binding assays revealed that the affinity of D95A mutant ribosomes for the CrPV IGR IRES element was decreased by ~1.6 fold as compared to wild-type ribosomes (Figure 1C). These studies were extended to mammalian cells to test whether the effect of Ψ modification impairment on ribosome ligand binding is evolutionarily conserved. Ribosomes isolated from hypomorphic Dkc1m MEFs, which show a significant reduction in total 18S and 28S Ψ levels (Figure S1), had ~2-fold lower affinity for CrPV IGR IRES than their wild-type counterparts (Figure 1D). Furthermore, assembly of the 48S pre-initiation complex on the CrPV IGR IRES was reduced by ~50% in the Dkc1m MEFs as compared to the WT control cells (Figures 1E, 1F). These studies demonstrate that global Ψ defects affect ribosome recruitment to the CrPV IGR IRES during initiation of translation in both fungal and mammalian cells.
Growth defects and temperature sensitivity are genetic indicators of defects in ribosome function. Standard 10-fold dilution spot assays revealed that the D95A mutant promoted a severe slow growth phenotype that was exacerbated at high temperature (37°C) (Figure S2). One important function of the ribosome is to faithfully maintain the translational reading frame. Viral mRNA signals that abrogate this function by programming ribosomes to shift frames (programmed ribosomal frameshifting, PRF) have proven to be of tremendous utility as readouts of translation fidelity (Dinman and Berry, 2006). Bicistronic dual luciferase reporters were used to quantify changes in PRF directed by the L-A −1 PRF, or the Ty1 retrotransposable element +1 PRF signals in yeast cells (Figure 2A). The D95A mutation promoted a ~3.5 fold increase in L-A mediated −1 PRF compared to wild-type cells (7.3% versus 2.0%), while Ty1 promoted +1 PRF was also increased, but to a lesser extent (6.2% versus 4.7%) (Figure 2B). To extend these studies to mammalian cells, we transfected a bicistronic vector harboring the HIV-1 −1 PRF signal (Figure 2A) into human cells where DKC1 expression was down-regulated by siRNA (Figure S3). siRNA-mediated knockdown of DKC1 diminished Ψ rRNA levels (Figure S4) and HIV-1 frameshifting was stimulated compared to control cells (1.3-fold increase) (Figure 2C). Notably, we have recently identified a number of functional −1 PRF signals encoded by mammalian mRNAs (Belew et al., 2008); one signal is located in the CCR5 mRNA (Hammell et al., 1999), which encodes the co-receptor for HIV-1 (Dragic et al., 1996), while a second is included in the human and mouse mRNAs encoding IL-7RA. siRNA-mediated knockdown of DKC1 in human cells increased CCR5 promoted −1 PRF by 2.4-fold, and by 1.8-fold for the human IL7-RA frameshift element (Figure 2C). −1 PRF mediated by the mouse IL7-RA frameshift element was also significantly increased in cells from Dkc1m compared to WT mice (Figure 2D). Altogether, these results indicate that the role of rRNA modifications in the control of translation fidelity is conserved from yeast to mammalian cells.
The effect of decreased Ψ modification on translational fidelity during termination was monitored using a bicistronic assay (Figure 2A). Recognition of an in-frame stop codon by the D95A mutant occurred with a ~2-fold greater frequency than in WT ribosomes (Figure 2E). These findings suggest that Ψ defects alter ribosome fidelity resulting in multiple molecular impairments, i.e. defective reading frame maintenance and stop codon readthrough in yeast and human cells.
Changes in ribosome/tRNA interactions can affect translational fidelity (Dinman and Berry, 2006). Therefore, we sought to biochemically monitor the effects of Ψ depletion on the interaction of ribosomes with their tRNA ligands. A-site binding was measured by incubating two-fold serial dilutions of a pre-formed ternary complex consisting of [14C]Phe-tRNAPhe•eEF1A•GTP with ribosomes pre-bound to a poly(U) mRNA and deacylated tRNAPhe in the P-site. Binding of aa-tRNA to the ribosomal A-site was decreased ~1.6 fold, as reflected by higher KdS in the D95A mutant as compared to ribosomes purified from WT cells (Figure 3A). P-site binding was measured by addition of Ac-[14C]Phe-tRNAPhe, a peptidyl tRNA analog that occupies the P/P site of the ribosome, to ribosomes pre-bound to a poly(U) mRNA. The D95A mutant promoted a roughly 2.5-fold decrease in affinity for Ac-aa-tRNA binding to the ribosomal P-site (Figure 3B).
Translational inhibitors that specifically bind to the ribosome are useful tools for examining changes in ribosome function. Anisomycin, which interferes with the binding of aa-tRNA to the A-site (Grollman, 1967) and inhibits −1 PRF (Dinman et al., 1997), was used to probe for defects in this region of the ribosome. Similarly, sparsomycin, which interferes with the binding of the 3′ end of peptidyl-tRNA (Grollman, 1967) and stimulates −1 PRF (Dinman et al., 1997), was used to probe for changes in P-site function. Paromomycin was used to probe interactions at the decoding center on the small subunit involving the aa-tRNA (Vicens and Westhof, 2001). The D95A mutant promoted marked hypersensitivity to paromomycin and sparsomycin (Figure 3C). Importantly, anisomycin appeared to rescue the growth defect of D95A mutant cells (Figure 3C), suggesting that global inhibition of −1 PRF could be used to treat X-DC. Thus, loss of Ψ results in altered sensitivities to translational inhibitors by affecting the interactions between tRNAs and ribosomes at three distinct locations: the A-site, the P-site, and the decoding center. As tRNA and CrPV IGR IRES share similar structural features, these findings strongly suggest that Ψ modification of rRNA generally functions to optimize ligand binding to ribosomes.
Base modifications in rRNA are highly conserved and are associated with the functionally important domains of the ribosome, suggesting a significant contribution of Ψ for proper ribosome performance (Ofengand, 2002). However, the molecular and biochemical mechanisms by which Ψ residues impinge on control of protein synthesis remain poorly understood. Here, we characterized the molecular consequences of global Ψ inactivation from yeast to human cells, showing that global rRNA pseudouridylation defects affect ribosome-ligand interactions, conferring significant biological costs. Thus, the potential impact of changes in Ψ on the rRNA is potentially widespread.
Impaired ribosome-ligand interactions due to decreases in Ψ affect numerous aspects of translational fidelity in yeast and mammalian cells. The dramatic effects on −1 PRF as a consequence of Ψ inactivation (Figure 2) are consistent with the requirement for slippage of both A- and P-site tRNAs (Liao et al., 2010): decreased affinities for both tRNAs as a consequence of Ψ inactivation appear to have a multiplicative effect on frameshifting. In contrast, Ty1 directed +1 PRF only involves P-site tRNA slippage (Liao et al., 2008), thus accounting for the smaller effect of the D95A mutant in response to this signal (Figure 2B). Stop codon readthrough results from improper selection of an aa-tRNA at a termination codon. Decreased ribosomal affinity for aa-tRNA A-site binding promotes increased dissociation rates of near- and non-cognate aa-tRNAs from termination codons, resulting in decreased rates of aberrant amino acid incorporation at the UAA stop codon. Indeed, the observation that of decreased UAA readthrough in D95A cells supports the hypothesis that the defect is specific to RNA ligands, and not to proteinacious ligands such as eRF1/eRF3 (Figure 2E). Decreased rates of nonsense-suppression are also consistent with the paromomycin resistance phenotype of the D95A mutant (Figure 3C).
Another consequence of defects in ribosome-ligand interactions is impaired IRES-mediated translation. The CrPV IGR IRES element forms a complex tertiary structure, one domain of which assumes a tRNA-like conformation that interacts with the ribosomal P-site (Costantino et al., 2008). Reduced affinity of D95A mutant ribosomes for this IRES element is consistent with the decreased ability of mutant yeast cells to initiate translation from it (Figure 1B and 1C). Importantly, these effects were also evident in Dkc1m primary cells as seen by the decreased ability of Dkc1m ribosomes to bind the CrPV IGR IRES as well as in 48S pre-initiation complex formation (Figures 1D, 1E). Our findings suggest that even small changes at the biochemical level that perturb ribosome/ligand interactions may lead to more pronounced effects on translational control and phenotypic effects. While greater defects in tRNA binding do not result in lethality in E. coli (McClory et al., 2010; Zaher and Green, 2010), similarly large changes may not be tolerated in yeast because of the presence of the non-functional ribosome decay (NRD) apparatus present in eukaryotic cells (LaRiviere et al., 2006). This is also supported by the fact that mutations in trans-acting eukaryotic initiation factors, such a eIF1A, have been reported to strongly affect recruitment of initiator tRNA (Saini et al., 2010) or IRES elements (Otto and Puglisi, 2004; Ji et al., 2004) without affecting cell viability. Thus, large changes in affinities for RNA ligands in the large subunit itself are probably unobtainable, and the observed ~40% decreases in ribosome affinity for RNA ligands promoted by D95A mutant ribosomes likely approach the tolerable limits for the viability of this single celled microorganism. Importantly, while numerous point mutations within the DKC1 gene have been characterized in X-DC, complete deletions have never been identified, and may not be compatible with life.
It is highly significant that a decrease in ligand binding to the ribosome is conserved from mammalian cells to yeast. In addition, the manifestation of these defects, such as decreased IRES-mediated translation and increased PRF, are also conserved. Computational studies suggest that approximately 10% of chromosomally-encoded open reading frames may harbor −1 PRF signals (Belew et al., 2008). In addition, numerous IRES-containing mRNAs have been identified in mammalian genomes, many of which encode important proteins involved in key cellular processes including cell cycle and apoptosis (Graber and Holcik, 2007; Spriggs et al., 2005). Thus, deregulation of gene expression as a consequence of these biochemical impairments may contribute, at least in part, to the broad range of symptoms displayed in X-DC and HH patients. It should be noted that DKC1 encodes for a multifunctional protein (Jiang et al., 1993; Mitchell et al., 1999; Ma et al., 2005) and disruptions in its activity may have even more wide-ranging molecular and cellular consequences. The data reported in this study provide important insights into how Ψ directly affects ribosomal function in vivo and how defects in the process impair specific ribosome-ligand interactions that are evolutionarily conserved.
Isogenic cbf5Δ yeast strains expressing CBF5 or cbf5-D95A from low copy vectors were constructed using standard plasmid shuffle techniques and are listed in Table S1A. Sensitivity to translational inhibitors and to high and low temperatures was determined using standard 10-fold dilution spot or filter disc assays.
CrPV IGR IRES dual-luciferase reporter (pSRT215) was derived from pSRT209 (Landry et al., 2009). Bicistronic reporters harboring the yeast L-A virus −1 PRF signal, the Ty1 derived +1 PRF signal, or an in frame UAA codon were previously described (Harger and Dinman, 2003). The dual-luciferase reporter containing the HIV-1 −1 PRF signal was described elsewhere (Grentzmann et al., 1998). The −1 PRF signal from Homo sapiens CCR5 was amplified from pCMV-XL4 (pJD819) (Grentzmann et al., 1998) and H. sapiens IL-7RA −1 PRF signal was obtained using sequence specific oligonucleotides from human cDNA. Mammalian dual-luciferase reporters were transfected into either MEFs or HeLa cells using Lipofectamine 2000. siRNA knockdown experiments in HeLA cells employed either DKC1 or control scramble siRNA oligos. Translational recoding was measured as described (Jacobs and Dinman, 2004; Landry et al., 2009).
Determination of steady-state binding constants for tRNAs were performed as previously described with modifications (Meskauskas and Dinman, 2010; Landry et al., 2009). Briefly, lysates of mid-log phase yeast cells were clarified by low speed centrifugation and supernatants were chromatographically purified (Leshin et al., 2010). Remaining tRNAs were cleared from eluted ribosomes with pH neutralized puromycin. Ribosomes were purified in high salt buffer by ultracentrifugation through glycerol cushions, and subsequently washed and resuspended in low salt buffer. To monitor [14C]Phe-tRNAPhe binding to the A-site, ribosomes were pre-activated in binding buffer containing polyU and deacylated tRNAPhe. Ternary complex was pre-assembled using HPLC purified [14C]Phe-tRNAPhe, GTP and soluble ammonium sulfate fraction containing yeast elongation factors. Binding reactions containing ribosomes (33.33nM) and serial dilutions of ternary complexes (0 to 128 pmoles) were filtered through nitrocellulose filters. Similar assays to monitor binding of Ac-[14C]Phe-tRNAPhe to the ribosomal P-site were performed using pre-activated ribosomes incubated with serial dilutions of HPLC purified Ac-[14C]Phe-tRNAPhe and processed as described above.
Assays of CrPV IGR IRES binding were performed as previously described (Landry et al., 2009). Briefly, 40S ribosomal subunits were isolated from wild type and cbf5-D95A yeast strains, or to wild type and Dkc1m MEFs. CrPV IGR IRES filter binding assays were performed with purified 40S subunits at a range of concentrations of radiolabeled IRES RNA. Radioactivity was quantified for both the bound and the free using a scintillation counter. Kd values were calculated assuming one binding site in Grafit (Gecces Software).
Total analysis of rRNA Ψ levels was carried out by HPLC (Montanaro et al., 2006; Bellodi et al., 2010b) with some modifications. 18S and 28S rRNA species derived from siRNA treated HeLa, or mouse MEFs were gel purified, digested with nuclease P1, dephosphorylated and analyzed by HPLC.
Analysis of the 48S pre-initiation complex formation was performed as described (Costa-Mattioli et al., 2004). [32P]-CrPV IRES RNA was incubated with cytoplasmic extracts prepared from WT and Dkc1m MEFs. Incubation was carried out for 2 min at 30°C in the presence of translation initiation inhibitors GMP-PNP and cycloheximide. Extracts were run on 10–30% sucrose density gradient and fractionated. Radioactivity was quantified using a scintillation counter.
Other methods are described in detail in Supplemental Experimental Procedures.
We thank John Carbon for the gifts of yeast strains (YTK227 and YCC133) and plasmids (pCBF5-BFG and pcbf5D65A-BFG). Many thanks to: Rasa Rakauskait, Trey Belew, and Michael Rhodin for their assistance; Maria Barna and Kim Tong for discussion and editing. This work was supported by a grant to JDD from the Public Health Service (5R01GM058859; 3R01GM058859-10A1S1), to SRT from the National Institutes of Health (R01GM084547; 3R01GM084547-01A1S1) and to DR from the National Institutes of Health (R01HL085572; 3R01HL085572-05S1). DR is a Leukemia & Lymphoma Society Research Scholar. CB is a Leukemia & Lymphoma Society (LLS) and Aplastic Anemia & Myelodysplastic Syndromes (MDS) International Foundation (AA&MDS IF) research fellow.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.