Transcriptome-wide mapping of pseudouridines: pseudouridine synthases modify specific mRNAs in S. cerevisiae

Abstract

We developed a novel technique, called pseudouridine site identification sequencing (PSI-seq), for the transcriptome-wide mapping of pseudouridylation sites with single-base resolution from cellular RNAs based on the induced termination of reverse transcription specifically at pseudouridines following CMCT treatment. PSI-seq analysis of RNA samples from S. cerevisiae correctly detected all of the 43 known pseudouridines in yeast 18S and 25S ribosomal RNA with high specificity. Moreover, application of PSI-seq to the yeast transcriptome revealed the presence of site-specific pseudouridylation within dozens of mRNAs, including RPL11a, TEF1, and other genes implicated in translation. To identify the mechanisms responsible for mRNA pseudouridylation, we genetically deleted candidate pseudouridine synthase (Pus) enzymes and reconstituted their activities in vitro. These experiments demonstrated that the Pus1 enzyme was necessary and sufficient for pseudouridylation of RPL11a mRNA, whereas Pus4 modified TEF1 mRNA, and Pus6 pseudouridylated KAR2 mRNA. Finally, we determined that modification of RPL11a at Ψ -68 was observed in RNA from the related yeast S. mikitae, and Ψ -239 in TEF1 mRNA was maintained in S. mikitae as well as S. pombe, indicating that these pseudouridylations are ancient, evolutionarily conserved RNA modifications. This work establishes that site-specific pseudouridylation of eukaryotic mRNAs is a genetically programmed RNA modification that naturally occurs in multiple yeast transcripts via distinct mechanisms, suggesting that mRNA pseudouridylation may provide an important novel regulatory function. The approach and strategies that we report here should be generally applicable to the discovery of pseudouridylation, or other RNA modifications, in diverse biological contexts.

Figure 1. Overview of Pseudouridine Site Identification Sequencing (PSI-Seq).

Chemical structures of pseudouridine (A) and CMCT (1-cyclohexyl-(2-morpholinoethyl)carbodiimide metho-p-toluene sulfonate) (B) are shown, as well as the CMC-adduct (C) generated by reaction of pseudouridine with CMCT followed by alkaline treatment. The bulky CMC group can stop reverse transcription of RNA. (D) Diagram of library preparation strategy for sequencing assay to identify pseudouridines in cellular RNA. The RNA sample is fragmented by alkaline hydrolysis and size-selected for RNA pieces 100 to 300 nt in length. RNA fragments are then reacted with CMCT to produce CMC-adducts (shown as a black hexagon). In parallel, the same sample is incubated with buffer lacking CMCT (mock treatment) as a negative control. RNA is then alkaline treated to remove adducts from non-pseudouridines. Linkers are ligated onto the 3′-end of the RNA fragments to allow for reverse transcription. Reverse transcription from the linkers is performed and cDNA is isolated. Size selection for truncated cDNA products with insert lengths of 20-80 nt is performed, yielding only cDNAs whose reverse transcription stopped before the end of the RNA fragment. These cDNAs are converted into sequencing libraries using a protocol adapted from the ribosome profiling technique (Ingolia et al. 2010).

Figure 2. PSI-Seq Identifies Known Pseudouridine Sites in Yeast Ribosomal RNA.

For each nucleotide in 18S and 25S rRNA, the log2 ratio of the number of reads (plus one) stopping at that position in the +CMCT library versus the –CMCT (mock) library was plotted. The dashed blue line indicates the cutoff for the top 60 sites in each condition, which includes all 43 known pseudouridine sites (red squares) that are not further modified. Sites among the top 60 that were located exactly one away from known pseudouridines, representing “secondary stops”, are marked with green circles. The data are plotted for yeast RNA isolated from both log phase (A–B) and heat shock (C) conditions. Scatterplots of the log₂ ratios shown in (A–C) are displayed for pairwise comparisons to assess reproducibility of values for known pseudouridines (red) and all sites (black) between independent experiments (D–F).

Figure 3. PSI-Seq Analysis of the Yeast Transcriptome Reveals mRNA Pseudouridylation Sites.

(A–B). For RNA isolated from yeast growing in log phase (A–B) or heat shock (C) conditions, pseudouridylation “scores” were calculated based on a regression analysis of the log2 transformed normalized read densities for each nucleotide position in treated (+CMCT) versus mock (-CMCT) samples (details in text). Values plotted on the x-axis indicate scores (+CMCT/−CMCT) from PSI-seq analysis of natural RNA samples. Values plotted on the y-axis indicate the difference between the natural RNA (+CMCT/−CMCT) scores and the corresponding score values obtained from PSI-seq analysis of unmodified in vitro transcribed control RNA samples. Each data point corresponds to a single nucleotide position within the transcriptome. Vertical and horizontal dashed blue lines illustrate the cutoff thresholds used as criteria for candidate pseudouridine identification for each experiment. Unique non-rRNA-derived uridine sites in the upper-right quadrant were classified as candidate mRNA pseudouridines (blue circles). Known pseudouridine sites from the ribosomal RNA are also shown (red circles).

Figure 4. Primer Extension Analysis of Pseudouridine Sites in RPL11a and TEF1 mRNAs.

Reverse transcription using fluorescent gene-specific primers for RPL11a (A–B) or TEF1 (C–D) was performed on RNA samples following treatment with CMCT (+CMCT) or mock reactions (-CMCT), and analyzed by capillary electrophoresis. At each position, the difference in primer termination rates between +CMCT and –CMCT conditions was computed from electropherogram intensities and is displayed in a gel-like format. Positions appearing with darker intensities correspond to sites with specific termination induced by CMCT treatment, as expected for pseudouridines. RNA samples examined were natural RNA isolated from wild-type yeast (wt), in vitro transcribed RNA as an unmodified control (IVT), or natural RNA isolated from a mutant yeast strain (pus1Δ or pus4Δ). The site of pseudouridylation is marked by an arrow. Quantification of the site-specific termination rate for each RNA sample is plotted for RPL11a psi-68 (B) and TEF1 psi-239 (D).

Figure 5. RPL11a RNA and TEF1 RNA Can Be Pseudouridylated by Cell Extract and Purified Pseudouridine Synthases.

A. Unmodified in vitro transcribed RPL11a RNA was either left untreated (IVT) or was incubated with cell extract made from wild-type yeast (wt) or mutant yeast lacking an individual pseudouridine synthase gene (pus1-pus6). The RNA was then analyzed by reverse transcription primer extension with or without CMCT treatment, and the data were plotted as described in figure 4. B. For these modification experiments, the proportion of reverse transcription that stopped at the site of pseudouridylation was plotted for both +CMCT (blue) and mock (red) conditions as described in figure 4B. C. In vitro transcribed RPL11a RNA was modified by purified pseudouridine synthase enzymes Pus1p through Pus8p, or untreated (IVT). The CMCT-RT stop experiment was performed, and the data were plotted as described in figure 4A. D. For these experiments, the proportion of reverse transcription that stopped at the site of pseudouridylation was plotted for both +CMCT (blue) and mock (red) conditions as described in figure 4B. E-H. Same as A-D, for TEF1 in vitro transcribed RNA.

Figure 6. Determination of Sequence Requirements for Pseudouridylation of RPL11a and TEF1 mRNAs.

A. 55 nt RNA constructs containing 55 nt (R55), 20 nt (R20), 10 nt (R10), or 6 nt (R6) of sequence corresponding to the sequence around the site of modification in RPL11a mRNA were in vitro transcribed and modified with purified Pus1p. The RNA was then used for the CMCT-RT stop experiment, and the data were plotted as described in figure 3A. B. For these modification experiments, the proportion of reverse transcription that stopped at the site of pseudouridylation was plotted for both +CMCT (blue) and mock (red) conditions as described in figure 3B. C. The sequences of the constructs used. For R20, R10, and R6, the sequences conserved from RPL11a are in blue. D. 55 nt constructs of containing the 55 nt of sequence surrounding the site of modification in RPL11a mRNA were in vitro transcribed with the following mutants: no mutation (R55), site of pseudouridylation (U68) to C (R55_0C), U69A (R55_1A), UCU(65-67) to AGC (R55_AGC) or both U69A and UCU->AGC (R55_AGC1A). The RNA was then used for the CMCT-RT stop experiment, and the data were plotted as described in figure 3A. E. For these modification experiments, the proportion of reverse transcription that stopped at the site of pseudouridylation was plotted for both +CMCT (blue) and mock (red) conditions as described in figure 3B. F. The sequences of the constructs used. The mutated nucleotides are in red. G-L. Same as A-F, for the following TEF1 constructs: 55 nt around the site of pseudouridylation in TEF1 (T55), 20 nt around the site (T20), 6 nt around the site (T6), 6 nt around the site in a 5 nt stem loop (T6-ds), T6ds with the U that is modified mutated to C (T6_ds_0C), the C 1 nt downstream mutated to U (T6_ds_1U), the G 2 nt upstream mutated to A (T6_ds_-2A) or both those mutations (T6_ds_1U-2A). For the sequence of T6-ds, the complementary nucleotides comprising the 5 nt stem are underlined.

Figure 7. Pseudouridylation of RPL11a and TEF1 mRNAs is Conserved in Other Fungal Species.

A,B. The CMCT-Stop experiment was performed on the orthologues of RPL11a in Saccharomyces mikitae and Schizosaccharomyces pombe. The proportion of reverse transcription that stopped at the site of pseudouridylation was plotted for both +CMCT (blue) and mock (red) conditions as described in figure 3A and B. C,D. As in A,B, for the orthologues of TEF1 in S. mikitae and in S. pombe.