|Home | About | Journals | Submit | Contact Us | Français|
Isomerization from uridine to pseudouridine (pseudouridylation) is largely catalyzed by a family of small ribonucleoproteins called box H/ACA RNPs, each of which contains one unique small RNA—the box H/ACA RNA. The specificity of the pseudouridylation reaction is determined by the base-pairing interactions between the guide sequence of the box H/ACA RNA and the target sequence within an RNA substrate. Thus, by creating a new box H/ACA RNA harboring an artificial guide sequence that base-pairs with the substrate sequence, one can site-specifically introduce pseudouridines into virtually any RNA (e.g., mRNA, ribosomal RNA, small nuclear RNA, telomerase RNA and so on). Pseudouridylation changes the properties of a uridine residue and is likely to alter the role of its corresponding RNA in certain cellular processes, thereby enabling basic research into the effects of RNA modifications. Here we take a TRM4 reporter gene (also known as NCL1) as an example, and we present a protocol for designing a box H/ACA RNA to site-specifically pseudouridylate TRM4 mRNA. Disease-related mutation can result in early termination of translation by creating a premature termination codon (PTC); however, pseudouridylation at the PTC can suppress this translation termination (nonsense suppression). Thus, the experimental procedures described in this protocol may provide a novel way to treat PTC-related diseases. This protocol takes 10–13 d to complete.
Pseudouridine (ψ), the 5′-ribosyl isomer of uridine1, was once considered as the fifth ribonucleoside in addition to the four other canonical ribonucleosides, namely adenosine, guanosine, cytidine and uridine2. Compared with uridine, the most distinct feature of ψ is the presence of an extra hydrogen-bond donor at its N1 position (Fig. 1a), which allows ψ to be involved in inter- or intramolecular hydrogen bonding through a water bridge. The water-bridged intramolecular hydrogen-bonding network is formed between the water molecule and the N1 imino-hydrogen and the 5′ phosphate oxygen of ψ as well as the phosphate oxygen of the 5′ preceding nucleotide3. In addition, ψ has been shown to improve base-stacking interactions within RNAs4 and be capable of altering RNA structure5. Furthermore, ψ is more polar than uridine6. In short, the presence of ψ in RNA could profoundly influence the properties of the RNA molecule and the cellular process in which it participates.
Pseudouridylation is a naturally occurring process that occurs post-transcriptionally. In eukaryotic nuclei, a group of noncoding RNAs, termed box H/ACA guide RNAs, are responsible for directing site-specific pseudouridylation of ribosomal RNA (rRNA) and spliceosomal small nuclear RNA7,8. Structurally, box H/ACA RNAs assume a functional hairpin-hinge-hairpin-tail conformation9,10 (Fig. 1b). The conserved box `H' (5′-ANANNA-3′, where N is any nucleotide) resides in the hinge region, whereas the box `ACA' (5′-ACA-3′) is located in the tail region just three nucleotides upstream of the 3′ end11. Within each hairpin, there is a single-stranded internal loop termed the pseudouridylation pocket, consisting of two discontinuous tracts of guide sequences (g1 and g1′, and g2 and g2′) that provide pseudouridylation-site specificity (Fig. 1b). Indeed, the guide sequences recognize the substrate RNA by base-pairing interactions12,13. Notably, the uridine residue to be pseudouridylated is always positioned between the two tracts of guide sequences and remains unpaired. The pseudouridylation site is usually 14–16 nt upstream of either box H or box ACA.
By constructing artificial guide RNAs, we recently demonstrated that site-specific modification can be introduced post-transcriptionally into a variety of RNA molecules, including mRNAs14–16, spliceosomal small nuclear RNAs17, rRNAs and telomerase RNA18. The ability to manipulate RNA modification allows us to probe RNA structure, identify important nucleotides and study RNA function.
Translation of an mRNA terminates when the ribosome encounters a nonsense/termination codon (UAA, UGA or UAG) in an mRNA19–21. Termination at one of these three nonsense codons yields a functional full-length protein. However, genetic mutations have been shown to generate a PTC within genes implicated in a large number of human diseases, including muscular dystrophy, cystic fibrosis and various types of cancer. When translation terminates at a PTC, a nonfunctional C-terminally truncated protein is produced, potentially leading to deleterious effects on cell survival and organ function22. Approximately one-third of human genetic diseases may be associated with PTCs23. Therefore, developing a protocol to specifically attenuate undesirable translation termination is attractive to both the academic and clinical communities.
The fact that a uridine residue appears in the first position of all three nonsense codons prompted us to investigate the possibility of whether introducing pseudouridylation into a PTC would result in nonsense suppression. To test this idea, we took advantage of the yeast Saccharomyces cerevisiae system and the well-established yeast box H/ACA guide RNAs24. Specifically, we generated a plasmid-borne PTC-containing TRM4 mRNA as a substrate and an artificial box H/ACA RNA (Fig. 1b) to target the PTC for pseudouridylation. Our data have shown that the uridine of this PTC is indeed pseudouridylated and that the pseudouridylated PTC triggers translational read-through (nonsense suppression) to produce a full-length Trm4 protein in yeast. Mass spectrometry indicated that the amino acids inserted at pseudouridylated PTCs are either serine/threonine (for ψAA and ψAG) or tyrosine/phenylalanine (for ψGA)25. Specifically, although ψAA codes for serine and threonine almost equally well, ψAG codes mostly for serine (90%) and far less efficiently for threonine (10%). On the contrary, ψGA codes mostly for tyrosine (80%) and less efficiently for phenylalanine (20%). Currently, it is not clear as to why specific amino acids are used to decode a particular pseudouridylated PTC. Nonetheless, our data suggest that RNA-guided pseudouridylation of RNA may represent a novel approach for inducing nonsense suppression in vivo. Although insertion of the amino acids (serine/threonine or tyrosine/phenylalanine) at the pseudouridylated PTCs could change the property of the protein, the benefit of producing a full-length protein, although not exactly a wild type (WT), often outweighs the harm that a single amino acid mutation brings. Indeed, a successful rescue of protein function through nonsense suppression was observed (judged by growth phenotype)25.
With respect to codon alterations, this approach also offers the possibility to study the potential expansion of the genetic code. As specific amino acids (serine/threonine or tyrosine/phenylalanine) can be incorporated at pseudouridylated PTCs, it is likely that pseudouridylation of sense codons will result in changes of coding specificity as well. In the canonical genetic code, over 50% of codons contain uridines. In theory, any of these codons can be targeted for pseudouridylation and codon alteration.
Our artificial box H/ACA guide RNAs were all derived from a naturally occurring box H/ACA RNA (snR81) with only guide sequences being altered, thus affording a simple and straightforward method. Notably, given the large number of patients who have PTC-related genetic diseases, such as cystic fibrosis, Duchenne muscular dystrophy, β-thalassemia and Hurler syndrome, this strategy holds promise for clinical applications as well.
One relatively major limitation of this protocol is the low efficiency of nonsense suppression in yeast cells. On the basis of western blot analysis, the efficiency of nonsense suppression is currently about 5–10% (read-through product/WT product). In our experience, the efficiency of nonsense suppression directly correlates with the efficiency of PTC pseudouridylation. Thus, our future goal is to improve the efficiency of mRNA pseudouridylation. This can probably be achieved through manipulating the box H/ACA RNAs used to guide ψ formation.
To evaluate the potential power of pseudouridylation in nonsense suppression, we first must generate a PTC-containing mRNA that is stably expressed in yeast. To this end, we chose a yeast 2μ plasmid, pTRM4-WT (Fig. 2), which encodes a C-terminally tagged Trm4p. The endogenous TRM4 gene encodes an S-adenosyl-l-methionine-dependent tRNA methyltransferase26. pTRM4-WT was a generous gift from E. Phizicky/E. Grayhack's laboratory. This system is ideal because it is capable of producing a large amount of C-terminally tagged Trm4p in yeast cells27, which will greatly facilitate the subsequent detection and purification of the nonsense suppression product.
Although all three nonsense codons (UAA, UAG, UGA) are available for targeting, here we take UAA as an example. For two reasons, we chose the phenylalanine at position 602 (F602) in Trm4p as the site for a PTC. First, the proximity of the F602 codon to the 3′ end of the gene maximized the stability of PTC-containing TRM4 mRNA, which would otherwise be degraded by nonsense-mediated mRNA decay22. Second, after tryptic digestion, F602 is located in the middle of a peptide fragment consisting of 12 aa residues (596–607; Fig. 3) that can be quantified by mass spectrometry, thus facilitating the identification of any amino acid encoded by the pseudouridylated PTC.
The introduction of a mutation in TRM4 mRNA at F602 was accomplished by using the site-directed mutagenesis kit from Stratagene together with pTRM4-WT and appropriate oligodeoxynucleotide primers (TRM4-F602X(TAA)-F1 and TRM4-F602X(TAA)-R1). As a result, the codon of F602 was changed from TTT to TAA, thus generating pTRM4-F602X(TAA). To target the PTC, an artificial TRM4-F602X(TAA)-specific box H/ACA guide RNA (based on the yeast naturally occurring snR81 box H/ACA RNA)24 was also constructed.
Construction of an artificial snR81 is facilitated by a four-primer overlapping PCR strategy. Note that as an alternative, a two-primer overlapping PCR or single oligonucleotide strategy, which require two long primers (~100 nt) or one overly long oligonucleotide (> 82 nt), respectively, could also be used to construct the artificial guide RNA. However, from a cost-savings perspective, it is desirable to use four-primer PCR (as four short oligonucleotides are much less expensive than one or two long oligonucleotides). Given the length of snR81 (182 nt) and the discontinuous guide sequences in each pseudouridylation pocket, it is convenient to design four overlapping pieces of oligodeoxynucleotide primers, each containing one guide sequence. Of the four pieces, there is one sense strand, snR81-TRM4-F602X(TAA)-F1, and three antisense strands, snR81-TRM4-F602X(TAA)-R1, snR81-TRM4-F602X(TAA)-R2 and snR81-TRM4-F602X(TAA)-R3. Within each primer, the guide sequences (g1, g1′, g2, g2′; see Figs. 1b and and4)4) were altered according to the target sequence (TRM4 mRNA) both at and flanking the PTC. Except for the guide sequences, all the other nucleotides of snR81 remain unchanged. Notably, the first (U) and second (A) residues of the PTC codon should be left unpaired by guide sequences to allow for pseudouridylation. In addition, using both pseudouridylation pockets of snR81 to target the PTC will enhance the modification efficiency.
In the four-primer overlapping PCR reaction, the four primers serve as templates for each other and are used consecutively to create an artificial snR81 (Fig. 4). The PCR reaction requires a different molar concentration for each of the four primers to efficiently generate the full-length snR81 (a 210-nt-long PCR product). Specifically, the overall relative ratio (in terms of pmol μl− 1) is 20:1:1:20 for the primers snR81-TRM4-F602X(TAA)-F1:snR81-TRM4-F602X(TAA)-R1:snR81-TRM4-F602X(TAA)-R2:snR81-TRM4-F602X(TAA)-R3. If the PCR products are heterogeneous, one can purify the band of correct size and then perform a second round of PCR using the purified band as template and snR81-TRM4-F602X(TAA)-F1 and snR81-TRM4-F602X(TAA)-R3 as primers. The PCR products are then digested with restriction enzymes and inserted into a high-copy Escherichia coli yeast shuttle plasmid, pSEC (snoRNA expression cassette, available on request)28, between BamHI and HindIII sites. The transcription of artificial snR81 is under the control of the Gal promoter, which turns on in the presence of galactose.
Similarly, in a control experiment, we designed four different oligodeoxynucleotide primers (snR81-Control-F1, snR81-Control-R1, snR81-Control-R2, snR81-Control-R3) to generate psnR81-Control, whose guide sequences were altered such that they have no targets within the yeast genome. Thus, psnR81-Control was used to serve as a guide-specificity control (Fig. 5 and Table 1).
The S. cerevisiae parental strain, BY4741, was used. Given the genotype of BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0), URA3 and LEU2 genes were chosen as markers for selecting pTRM4 and psnR81 variants, respectively. Two alternate yeast transformation strategies can be used: co-transformation and sequential transformation. For co-transformation, both the pTRM4 and psnR81 variants are transformed into competent yeast cells in a single experiment, and transformants are screened on a –Leu/–Ura double-dropout solid medium. Sequential transformation, however, involves two independent consecutive transformations of the pTRM4 and psnR81 variants. Practically, co-transformation yields relatively lower (but still acceptable) transformation efficiency but is much less laborious compared with sequential transformation. Thus, only the co-transformation procedure is described in this protocol.
To prepare the yeast crude extract, cells are grown to mid-log phase in SGal–Leu–Ura double-dropout liquid medium. Because we experienced loss of expression of the plasmid-borne gene owing to the relaxation of auxotrophic pressure, YPGal medium should not be used to grow transformed cells to prepare extract. The presence of 2% (wt/wt) galactose in the culture medium is essential for activating the Gal promoter in the pSEC plasmid, which drives expression of both the TRM4 and snR81 variants.
For small-scale preparation of yeast crude extract (≤1 ml of settled wet cells), the sterile acid-washed glass beads method is ideal (see PROCEDURE). For large-scale preparation, we recommend using a homogenizer to process 10–20 ml of settled wet cells and using a mortar and pestle or French press to process more than 20 ml wet cells.
To evaluate the expression level of Trm4p, monoclonal anti-protein A (Sigma-Aldrich) and anti-Eno1p (gift from M. Dumont) are used to visualize both Trm4p and Eno1p (as a loading control). Eno1p is the phosphopyruvate hydratase that catalyzes the conversion of 2-phosphoglycerate to phosphoenolpyruvate during glycolysis29. Although the two primary antibodies are compatible with each other, we recommend that the membrane be probed sequentially to greatly reduce the nonspecific bands and background. Other than Eno1p, many housekeeping genes in S. cerevisiae could also serve as loading controls. If hybridization is carried out at 4 °C, the primary antibody solution can be reused several times.
As mentioned in the INTRODUCTION, pTRM4-F602X(TAA) is the construct that allows the use of the C-terminal tag to visualize nonsense suppression (translational read-through) products. It is then expected that when the PTC is not pseudouridylated, the C-terminally truncated Trm4p (without the tag) will be produced as a result of premature translation termination. When the PTC is pseudouridylated, however, nonsense suppression is induced, thus leading to the production of full-length Trm4p and its C-terminal tag, which allows for the detection of the nonsense suppression product by western blotting.
Figure 5 shows the western blot analysis of extracts prepared from cells expressing TRM4-WT (nonsense-free) or TRM4-F602X(TAA). When cells were transformed with pTRM4-WT, a strong protein A-tag signal was detected (Fig. 5, lane 1). However, when cells were co-transformed with TMR4-F602X(TAA) and a guide RNA containing a control guide sequence, the protein A-tag signal was undetectable (Fig. 5, lane 3). In contrast, when cells were co-transformed with pTRM4-F602X(TAA) and psnR81-TRM4-F602X(TAA), a fairly strong protein A-tag signal was detected, indicating nonsense suppression (Fig. 5, lane 2).
We thank the members of the Yu laboratory for discussion and inspiration. We also thank E. Phizicky, E. Grayhack and M. Dumont (University of Rochester Medical Center) for providing some of the reagents used in this work, and F. Hagen for helping analyze mass spectrometry data.
AUTHOR CONTRIBUTIONS C.H., G.W. and Y.-T.Y. prepared the figures and wrote the manuscript.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.