Epitranscriptomic mapping of RNA modifications at single-nucleotide resolution using rhodamine sequencing (Rho-seq)

Abstract

The recent development of epitranscriptomics revealed a new fundamental layer of gene expression, but the mapping of most RNA modifications remains technically challenging. Here, we describe our protocol for Rho-Seq, which enables the mapping of dihydrouridine RNA modification at single-nucleotide resolution. Rho-Seq relies on specific rhodamine-labeling of a subset of modified nucleotides that hinders reverse transcription. Although Rho-Seq was initially applied to the detection of dihydrouridine, we show here that it is applicable to other modifications including 7-methylguanosine or 4-thiouridine. For complete details on the use and execution of this protocol, please refer to Finet et al. (2022).

Keywords: Gene Expression; Molecular Biology; RNAseq; Sequence analysis.

Graphical abstract

Figure 1

Overview of Rho-seq workflow (A) A chemical treatment is applied to total RNA extract to label dihydrouridine (D) with rhodamine. It begins with the specific reduction of D with sodium borohydride (NabH₄). Upon the addition of a nucleophile NH₂-fluorophore such as rhodamine in acidic condition, a covalent bond is formed with Schiff base intermediate tetrahydrocitidine (Kaur et al., 2011). Along with the effective rhodamine labeling condition (WT R⁺, highlighted in red), control condition include mock labeling (WT R^-) where KOH substitute for NaBH₄ and D is therefore not reduced (R^-, highlighted in orange), as well as dihydrouridine-free total RNA extracts obtained from a strain where each dihydrouridine synthase is deleted (Δ4dus R⁺ and R^- , highlighted in cyan blue). Importantly, other RNA modifications can be specifically labeled using the same strategy, including m⁷G and s⁴U. (B) R⁺ and R^- treated RNAs containing uridine only (U, in cyan blue) from the Δ4dus condition, dihydrouridine (D, in orange) from the WT R^- condition and rhodamine-tetrahydrocitidine (R, in red) from the WT R⁺ condition are then subject to library preparation for high throughput sequencing. After RNA fragmentation, an RNA adapter blocked in 3′ with dideoxycytidine (blue line ending with a filled circle in 3′) is ligated to the fragmented RNA. Then, DNA primers complementary to the RNA adapter prime reverse transcription reactions. Reverse transcription either ends at the end of the RNA fragments (dotted lines in cyan and orange) or can be stopped prematurely by the bulky rhodamine molecule in the WT R⁺ condition (dotted red line ending with a filled circle in 3′). Finally, after the ligation of a DNA adaptor (in cyan, also blocked in 3′) in 3′ of the cDNA, the library is amplified by PCR with primers complementary to the adapters flanked with additional barcoded sequences. The amplified library is finally sequenced according to Illumina® chemistry in paired-end. (C) After trimming and mapping the sequenced paired-end reads to a reference sequence (R1 and R2 refer to the left and right read of a pair), the D-ratio is calculated at a single nucleotide resolution for each condition. It is the ratio between the number of reverse transcription stop events – reflected by the number of R2 reads starting at a position – and the fragment coverage. To robustly identify putative D-sites, i.e., transcriptomic position where the D-ratio is significantly higher in the WT R⁺ condition while controlling for all other factors, a three-stage approach is applied. First, the sites are filtered based on unsupervised criterion independent of the test statistic (see main text for details). Then, the D-ratio of the remaining sites is modelized in a generalized linear model of the binomial family with a logit link where the treatment (R⁺ and R^-), the strain (WT and Δ4dus) and their interaction are explanatory variables. The p-value of the coefficient of the strain:treatment interaction is computed and corrected using the Benjamini-Hochberg procedure (FDR). Finally, the effect size (D-fold change) is calculated as the ratio between the average D-ratio of the WT R⁺ condition against the average D-ratio of all the other control conditions.

Figure 2

Application of Rho-seq beyond dihydrouridine (A) R⁺-dependent RT stop in S. pombe 18S rRNA. The proportion of RT-stop (D-ratio) is indicated along the 18S rRNA (SPRRNA.43). The positions 1208 and 1616 corresponding to two previously described modifications (m¹acp³ψ and m⁷G, respectively) are highlighted. (B) Distribution of the position of RT-stop sites on E.coli tRNAs dependent on the R⁺ treatment but independent from dus-a-b-c genes. The Rho-seq analytical pipeline was modified to assess the significance of the treatment effect only (we computed the wald test of the treatment factor instead of the strain:treatment interaction in step 16 of the statistical analysis). The distribution of the position of sites significantly affected by the R⁺ treatment peaks at position 8, a widespread s⁴U modification site conserved in prokaryotes. (C) R⁺-dependent RT stop in E. coli tRNA-Pro(GGG). The proportion of RT-stop (D-ratio) is indicated along the Pro(GGG) tRNA (b2189). The positions 8 and 20 corresponding to two previously described modifications (S⁴U and D, respectively) are highlighted. (D) R⁺-dependent RT stop in E. coli tRNA-Lys(UUU). The proportion of RT-stop (D-ratio) is indicated along the tRNA-Lys(UUU) (b0743). The positions 16, 20 and 35 corresponding to three previously described modifications (D, D and cmnm⁵s²U, respectively) are highlighted.

Figure 3

A dihydrouridylated spike-in internally controls Rho-seq (A) Cartoon representation of the 150 nucleotides (nt) long spike-in. The sequence (produced by in vitro transcription) contains a unique uridine (U) position, that is replaced by D when the spike-in is transcribed in presence of rDTP instead of rUTP. A radiolabeled RT primer can prime reverse transcription in primer extension experiments. (B) Primer extension assay on fully dihydrouridylated spike-in mixed with S. pombe total RNA before the R+ and R- treatment visualized on a 15% poly-acrylamide gel. The migration patterns of the full-length RT product (5′ end), of the premature transcription stop site at the unique D position (D-sites), and of the excess labeled RT primer are indicated. (C) D-ratio (green line) and fragment coverage (blue line and shaded area) along the fully dihydrouridylated spike-in sequence as determined by Rho-seq. In the R+ condition, a sharp drop in coverage is observed one nucleotide downstream the unique D (position 44, highlighted with a dotted red line), that coincide with a sharp increase in D-ratio.

Figure 5

Expected results (A) Comparative rhodamine intensity signal measured at 520 nm in a dot blot assay after rhodamine labeling (R⁺) or mock treatment (R^-) with WT and Δ4dus total RNA. Methylene blue staining serves as loading control. (B) Typical size distribution of samples at various step of the Rho-seq procedure as analyzed using bioanalyzer RNA pico for RNA samples or DNA high-sensitivity (HS) chips for cDNA samples. From top to bottom: total RNA before rhodamine labeling, total RNA after rhodamine labeling (the acidic treatment causes RNA degradation), ribodepleted RNA, and PCR-amplified rho-seq cDNA library. RIN = RNA integrity number. (C) Example Rho-seq results for a Lysine tRNA (SPMITTRNALYS.01). High D-ratio can be observed one nucleotide upstream of the known D-sites at position 16 and 20.

Figure 4

Rhodamine labeling workflow (A) Detailed workflow of the rhodamine labeling protocol. See main text for details. (B and C) Upon the addition of acetic acid to the NaBH₄-treated samples (b,), the solution reacts rapidly and froth (C). (D) The samples take an orange tint after the addition of rhodamine-110 in acidic conditions. (E) Covering a thermomixer in aluminum foil allows the incubation of the samples with rhodamine in the dark. (F) Adjusting the pH with tris and phenol turn the samples yellow. (G) Following phenol RNA purification, unincorporated rhodamine separates in the lower, organic phase, while RNA (including rhodamine-labeled RNA) separates in the upper, aqueous phase.