Unbiased screen of RNA tailing activities reveals a poly(UG) polymerase

Abstract

Ribonucleotidyl transferases (rNTases) add untemplated ribonucleotides to diverse RNAs. We have developed TRAID-seq, a screening strategy in Saccharomyces cerevisiae to identify sequences added to a reporter RNA at single-nucleotide resolution by overexpressed candidate enzymes from different organisms. The rNTase activities of 22 previously unexplored enzymes were determined. In addition to poly(A)- and poly(U)-adding enzymes, we identified a cytidine-adding enzyme that is likely to be part of a two-enzyme system that adds CCA to tRNAs in a eukaryote; a nucleotidyl transferase that adds nucleotides to RNA without apparent nucleotide preference; and a poly(UG) polymerase, Caenorhabditis elegans MUT-2, that adds alternating uridine and guanosine nucleotides to form poly(UG) tails. MUT-2 is known to be required for certain forms of RNA silencing, and mutants of the enzyme that result in defective silencing did not add poly(UG) tails in our assay. We propose that MUT-2 poly(UG) polymerase activity is required to promote genome integrity and RNA silencing.

Figure 1.. TRAID-Seq assay measures nucleotide addition activity in vivo.

(a, b) TRAID-Seq strategy. (a) tRNA^Ser(AGA) variable arm (gray) is mutated to an MS2 stem loop (cyan) to form the tRNA reporter. (b) Left, tRNA reporter is co-expressed with an MS2 coat protein-rNTase fusion in S. cerevisiae. The tethered rNTase adds nucleotides to the 3′ end of the tRNA. Right, RT-PCR analysis to detect A tails or U tails added by control rNTases, relative to empty vector or a no-reporter control. Lanes marked with a dash indicate reactions performed without reverse transcriptase. Representative gel image from four independent experiments. (c) Schematic of sample processing. (d, e) Tail-o-grams of nucleotides added by control rNTases, C. elegans PUP-2 (d) and C. elegans GLD-2 (e). Percent of each nucleotide at each tail length is color-coded and plotted on the left y-axis; U (green), C (yellow), G (purple), A (brown). Tails lengths of five nucleotides or greater are shown for clarity (see Online Methods). The number of tails detected per million heptamers (TPMH) are indicated by black diamonds and correspond to the log scale on the right y-axis.

Figure 2.. Analyses of nucleotide addition activities of 40 noncanonical rNTases from seven species.

Overall percentages of each nucleotide added by (a) H. sapiens, (b) C. elegans, and (c) fungal rNTases. (d) Categorization of rNTases as PUPs, PAPs, CCA-adding enzymes, or those with unique activities. rNTases are color-coded by organism. Gray boxes (top) indicate previously characterized (known) enzymes, and black boxes (bottom) indicate enzyme activities identified in this study (new).

Figure 3.. Nucleotide addition activity of S. pombe SPAC1093.04 and S. cerevisiae Cca1.

(a) Left, tail-o-gram depicting nucleotide composition in each added tail length added by S. pombe SPAC1093.04 and number of tails normalized to unique heptamer sequences. Right, most abundant tail sequences added to tRNA reporter containing a 3′ CC, or 3′ CCA end. (b) Left, tail-o-gram depicting nucleotide composition in each added tail length added by S. cerevisiae Cca1 and number of tails normalized to unique heptamer sequences. Right, most abundant tail sequences added to tRNA reporter containing a 3′ CCA end. (c) Sequence motif effect analysis of tails added by SpSPAC1093.04 (red, n=5) and ScCca1 (black, n=3). Each adjusted p-value quantifies the significance of contribution of the indicated oligonucleotide to the variation in tail sequence read counts. Significances for dinucleotide (CC) and trinucleotides (CCA) after multiplicity correction with the Bonferroni procedure are shown. A dashed line indicates significance level 0.05. The -log₁₀ p-values from left to right in the figure are 300, 148, 0.87, and 313. (d). cca1-1 mutant strains containing CEN plasmids expressing indicated plasmids were serially diluted, spotted on SD-Ura-Leu media and grown at 37°C for 3 days or 23°C for 4 days. This experiment was repeated twice with similar results.

Figure 4.. CeMUT-2 is a poly(UG) polymerase.

(a) Tail-o-gram depicting CeMUT-2 nucleotide addition activity in TRAID-Seq. (b) The most abundant tail sequences identified in two biological replicates of CeMUT-2 TRAID-Seq assays. (c) UG tail sequences from two biological replicates of CeMUT-2 activity in X. laevis oocytes. (d) Analysis of all possible dinucleotides in the tails added by CeMUT-2 from 8 independent biological replicates. A heatmap of -log₁₀ p-values for individual dinucleotides is shown. Each p-value quantifies the significance of adjusted contribution of each dinucleotide to the variation in tail sequence read counts. Dinucleotides with a significant effect after multiplicity correction at significance level 0.05 are marked with an asterisk (*). Details of statistical analyses performed are provided in the Online Methods (Computational Analyses of Sequence Motifs).

Figure 5.. CeMUT-2 mutants defective for RNAi lack poly(UG) polymerase activity.

(a) Schematic of CeMUT-2 isoforms and tested mutations, known catalytic mutants (pink), and mutants identified in forward genetic screen (blue). NTD, Nucleotidyl transferase domain; PAPd, Poly(A) polymerase-associated domain. * indicates that a truncated version of CeMUT-2 was made to recapitulate this nonsense mutant. (b) Percent of nucleotides added by each CeMUT-2 enzyme variant. Percent of tails containing UG repeats, standard deviation, and number of biological replicates are indicated. (c) Model depicting potential roles of poly(UG) tails in small RNA amplification in C. elegans. Poly(UG) tails could directly recruit RNA-dependent RNA polymerase (RdRP) (left). Alternatively, poly(UG) tails could be identified by a poly(UG) binding protein (UG-BP), which then recruits RdRP (right). In both cases, the UG tails could be single-stranded or form a higher-order structure.