A systematic dissection of determinants and consequences of snoRNA-guided pseudouridylation of human mRNA

RNA can be extensively modified post-transcriptionally with >170 covalent modifications, expanding its functional and structural repertoire. Pseudouridine (Ψ), the most abundant modified nucleoside in rRNA and tRNA, has recently been found within mRNA molecules. Due to technical challenges, it remained unclear whether pseudouridylation of mRNA can be snoRNA-guided, which has important implications for understanding the physiological target spectrum of snoRNAs and for their potential therapeutic exploitation in genetic diseases. Here, using a massively parallel reporter based strategy we simultaneously interrogate Ψ levels across hundreds of synthetic constructs with predesigned complementarity against endogenous snoRNAs. Our results demonstrate that snoRNA-mediated pseudouridylation can occur on mRNA targets. However, this is typically achieved at low efficiencies, and is constrained by the localization of mRNA, by the expression levels of snoRNAs and by the length of the snoRNA:mRNA complementarity stretches. We exploited these insights for the design of snoRNAs targeting pseudouridylation at premature termination codons, which was previously suggested to suppress translational termination. However, in contrast to previous studies, in this and follow-up experiments we observe no evidence for readthrough of pseudouridylated stop codons. Our study enhances our understanding of the scope, ‘design rules’ and constraints of snoRNA-mediated pseudouridylation, and challenges a key functional outcome associated with this modification. Experiment #1: 2 samples of Pol-I- and 2 samples of Pol-II-driven libraries transfected into HEK-293T cells, each treated with CMC or input. Extracted RNA underwent targeted psi-seq to determine pseudouridine level of constructs (used to create Table S2). Experiment #2: three samples of Pol-II-driven libraries co-transfected with ACA21 and ACA61 overexpressing plasmids, or with a control (empty) overexpression plasmid or with no additional vectors. Each sample was treated with CMC. all 3 samples underwent targeted psi-seq (used in Table S2). overexpression and empty plasmid samples also underwent RNA-seq to determine abundance of snoRNA (used to create Table S5). Experiment #3: two samples of Pol-II-driven libraries co-transfected with either control or DKC1-targeting siRNAs, each treated with CMC or input. All 4 samples underwent targeted psi-seq to determine pseudouridine levels of construct (used to create Table S2) and RNA-seq to determine abundance of DKC1 and snoRNAs (used to create Table S4). The siControl sample treated with CMC also underwent psi-seq to determine pseudouridine level of endogenous sites in rRNA (used to create Table S6). Experiment #4: a single sample from either HEK-293T, A549, MCF7 or K562 was transfected with a Pol-II-driven library and treated with CMC or input. Samples underwent both targeted psi-seq and RNA-seq to determine pseudouridine level of constructs (used to create Table S2) and abundance of snoRNAs (used to create Table S7). Experiment #5: total RNA that was extracted from HEK293T samples are used to generate cDNA libraries using a psi-seq protocol to assess pseudouridine levels in endogenous rRNA targets (used to create Table S6). Experiment #6: DNA reporters harboring a single thymidine were in vitro transcribed in the presence of uridine or pseudouridine to assess the effect of bases downstream to the modified site on psi-rate measured by targeted psi-seq (used to create Table S3). Experiment #7: plasmids expressing a bi-cistronic fluorescent reporter separated by a sequence element including a premature stop codon from the DMD gene were transfected into HEK-293T cells alongside a WT or modified sno61 construct targeting the DMD stop codon for pseudouridylation. Purified RNA was pooled, ribo-zero treated and was used to generate a pooled psi-seq library (used to create Figure 5b).