Inverse-folding design of yeast telomerase RNA increases activity in vitro

Abstract

Saccharomyces cerevisiae telomerase RNA, TLC1, is an 1157 nt non-coding RNA that functions as both a template for DNA synthesis and a flexible scaffold for telomerase RNP holoenzyme protein subunits. The tractable budding yeast system has provided landmark discoveries about telomere biology in vivo , but yeast telomerase research has been hampered by the fact that the large TLC1 RNA subunit does not support robust telomerase activity in vitro . In contrast, 155-500 nt miniaturized TLC1 alleles comprising the catalytic core domain and lacking the RNA's long arms do reconstitute robust activity. We hypothesized that full-length TLC1 is prone to misfolding in vitro . To create a full-length yeast telomerase RNA predicted to fold into its biological relevant structure, we took an inverse RNA folding approach, changing 59 nucleotides predicted to increase the energetic favorability of folding into the modeled native structure based on the p-num feature of Mfold software. The sequence changes lowered the predicted ∆G in this "determined-arm" allele, DA-TLC1, by 61 kcal/mol (-19%) compared to wild type. We tested DA-TLC1 for reconstituted activity and found it to be ∼5-fold more robust than wild-type TLC1, suggesting that the inverse-folding design indeed improved folding in vitro into a catalytically active conformation. We also tested if DA-TLC1 functions in vivo and found that it complements a tlc1 ∆ strain, allowing cells to avoid senescence and maintain telomeres of nearly wild-type length. However, all inverse-designed RNAs that we tested had reduced abundance in vivo . In particular, inverse-designing nearly all of the Ku arm caused a profound reduction in telomerase RNA abundance in the cell and very short telomeres. Overall, these results show that inverse design of S. cerevisiae telomerase RNA increases activity in vitro , while reducing abundance in vivo . This study provides a biochemically and biologically tested approach to inverse-design RNAs using Mfold that could be useful for controlling RNA structure in basic research and biomedicine.

Figure 1.. Determined-arm DA-TLC1 RNA: inverse-designed to fold more stably than wild-type TLC1 into the biologically relevant secondary structure.

Shown are the lowest-free-energy Mfold P-num computational secondary structure predictions for wild-type and Determined-Arm TLC1 (DA-TLC1). (A) Mfold secondary structure model for wild-type TLC1. Nucleotides are colored according to the P-num feature of the Mfold web server, which conveys the “determinedness” of low-free energy RNA folding predictions, with the most well-determined nucleotides on the red end of the ROYGBIV spectrum. The initial ΔG of folding of wild-type TLC1 was predicted by Mfold to be −321 kcal/mol. (B) Lowest-free energy Mfold secondary model for DA-TLC1, with nucleotides colored in the P-num format as in A. The determined Ku arm was designed to fold into the phylogenetically derived structure for the wild-type arm. The initial ΔG of folding was predicted to be −382 kcal/mol.

Figure 2.. DA-TLC1 RNA can function in place of wild-type TLC1 in vivo, maintaining slightly shorter telomeres and exhibiting reduced abundance.

(A) Like wild-type TLC1, DA-TLC1 allows yeast cells to grow perpetually and avoid senescing. Each telomerase RNA allele was harbored on a centromere-containing plasmid in a tlc1Δ rad52Δ yeast strain, and restreaked for over 250 generations. Growth shown at 250 generations (10 restreaks on plates). The tlc1Δ strains senesce by ~125 generations. (B) Telomeres in DA-TLC1 cells are ~23 bp shorter than wild-type cells. Shown is a Southern blot of XhoI-digested genomic DNA probed with telomeric repeat sequences. Change in telomere length relative to wild-type TLC1 is indicated for two isolates, ± S.D. (C) The abundance of DA-TLC1 RNA is reduced compared to wild-type. Two independent isolates at 250 generations are shown. Northern blot showing reduced DA-TLC1 RNA levels compared to wild-type. (D) Dot blot of total cellular RNA probed for the shared 3′ end of TLC1 (upper panel), which is common to all RNA variants, and normalized to U1 snRNA control spots (lower panel). Normalized telomerase RNA abundance relative to TLC1 from four isolates is indicated, ± standard error. Dot blots are required to study TSA-T since this RNA allele does not denature in urea-PAGE gels used for northern blotting (as in C) due to its long dsRNA arms.

Figure 3.. DA-TLC1 telomerase activity is increased compared to wild-type TLC1.

The indicated RNAs were synthesized in a rabbit reticulocyte lysate transcript-translation system along with Est42 (TERT), immunopurified and assayed for activity on a telomeric primer in the presence of [alpha-32P]-dGTP. Activity quantification is normalized to internal loading control and is the average ± standard deviation relative to Mini-T activity.

Figure 4.. Individual inverse-designed arms each contribute to increased DA-TLC1 activity in vitro and decreased RNA abundance in vivo.

(A) Mfold P-num models for the Ku arm of wild-type, DK-TLC1, and DPhyK-TLC1 alleles. (B) DA-TLC1 supports telomerase function in an in vitro reconstituted telomerase assay. As in Figure 3, TLC1 variants were co-expressed with ProA-Est2 in an in vitro transcription and translation system, co-immunopurified, and reacted with a telomeric DNA primer. Averages ± standard deviation of 2–4 activity-assay replicates of each enzyme preparation are shown. (C) Cells expressing Determined-Arm TLC1 variants grow well and do not senesce. Telomerase RNA alleles were expressed from CEN plasmids in a tlc1Δ, rad52Δ strain. Growth at 250 generations is shown. (D) Telomere Southern blot showing that DA-, DK-, and DET-TLC1 alleles support telomeres near wild-type length, whereas DPhyK-TLC1 and -DA-TLC1 exhibit very short telomeres. (E) Northern blot showing that whereas most inverse-designed TLC1 alleles have modestly reduced RNA abundance, DPhyK alleles have very low levels.