A New Specific and Sensitive RT-qPCR Method Based on Splinted 5' Ligation for the Quantitative Detection of RNA Species Shorter than microRNAs

Abstract

Recently, we discovered a new family of unusually short RNAs mapping to 5.8S ribosomal RNA (rRNA) and which we named dodecaRNAs (doRNAs), according to the number of core nucleotides (12 nt) their members contain. To confirm these small RNA-sequencing (RNA-Seq) data, validate the existence of the two overly abundant doRNAs-the minimal core 12-nt doRNA sequence and its + 1-nt variant bearing a 5' Cytosine, C-doRNA-and streamline their analysis, we developed a new specific and sensitive splinted 5' ligation reverse transcription (RT)-quantitative polymerase chain reaction (qPCR) method. This method is based on a splint-assisted ligation of an adapter to the 5' end of doRNAs, followed by RT-qPCR amplification and quantitation. Our optimized protocol, which may discriminate between doRNA, C-doRNA, mutated and precursor sequences, can accurately detect as low as 240 copies and is quantitatively linear over a range of 7 logs. This method provides a unique tool to expand and facilitate studies exploring the molecular and cellular biology of RNA species shorter than microRNAs.

Keywords: 5′ ligation; RT-qPCR; adapter; methods for functional RNA studies; quantification; small RNA; splint.

Figure 1

Schematic representation of the splinted 5′ ligation RT-qPCR method established to detect doRNAs. In Step 1, the 5′ RNA adapter was pre-annealed with the 4:2 splint DNA. In Step 2, the 5′ adapter:splint complex was added to total RNA to target the adapter ligation to the doRNA and the C-doRNA by base pairing; the splint specified the adapter ligation to the 5′ end of the doRNA or C-doRNA sequence. The ligation of the adapter to the doRNA or C-doRNA, templated by the splint, required the 5′ phosphate (marked as an encircled P) of the doRNA or C-doRNA, and was mediated by the T4 RNA ligase. The resulting adapter-ligated RNA intermediate (Step 3) was reverse transcribed by adding a poly(A) tail, and cDNA was synthesized using a poly(T) primer with a 3′ degenerate anchor and a 5′ universal tag (Step 4). Finally, doRNA and C-doRNA, 5′ extended by the adapter, were quantitated by qPCR using a set of locked nucleic acid (LNA)-modified oligonucleotides, with the forward primer specific to the RNA, and the reverse primer specific to the universal tag (Step 5).

Figure 2

Optimization of the DNA splints. Splints of various length were assessed for their ability to optimize 5′ ligation RT-qPCR detection of doRNA and C-doRNA. Cycle quantitation (Cq) data were normalized on U6 RNA, and the fold changes were calculated versus the optimal 4:2 splint condition (mean ± SEM; n = 3 independent experiments). Letters above the bars represent conditions with a normalized fold change significantly different from the control (without splint,-splint): A, p < 0.0097 (two-way ANOVA with Holm–Sidak’s post hoc test).

Figure 3

Assessment of the importance of the main reaction medium components. The importance of the essential components of the splinted 5′ ligation buffer was assessed by quantitatively evaluating doRNA and C-doRNA detection in their absence (−) or presence (+). Cycle quantitation (Cq) data were normalized on U6 RNA, and the fold changes were calculated versus the optimized buffer condition (control) containing all the components (mean ± SEM; n = 3 independent experiments). The 4:2 splint was used in this experiment. Letters above the bars represent conditions with a normalized fold change significantly different from the control: A, p < 0.0260; B, p < 0.0399 (two-way ANOVA with Holm–Sidak’s post hoc test).

Figure 4

Amplification of doRNA and C-doRNA by splinted 5′ ligation RT-qPCR was specific. (A) Melting curve of the qPCR reaction products following splinted 5′ ligation RT-qPCR detection of doRNA (in dark blue) or C-doRNA (in light blue) in total RNA samples from cultured N2a cells. RFU, relative fluorescence unit. (B) Amplification of a single product upon splinted 5′ ligation RT-qPCR detection of RNA and C-doRNA. qPCR products were separated by 20% acrylamide/8 M urea gel electrophoresis and visualized by nucleic acid staining (RedSafe). (C) DNA contained in the bands excised from lanes 3 and 4 was extracted and cloned into the pBluescript II plasmid. DNA sequencing confirmed the identity of the expected amplified product (in color). UP, universal primer.

Figure 5

Splinted 5′ ligation RT-qPCR may discriminate between doRNA, C-doRNA and precursor sequences. (A) Synthetic oligonucleotide sequences used to test the specificity of our splinted 5′ ligation RT-qPCR method. Mutated C-doRNA was obtained by substituting G2 to U2 near the 5′ end. The 5′ and 3′ extended C-doRNAs were 10-nt extensions of C-doRNA in the 5′ or 3′ direction of the 5.8S rRNA precursor sequence, respectively. (B,C) Total RNA (300 ng) from N2a cells was used, followed by addition of 6.02 × 10⁹ copies of the indicated synthetic oligonucleotides (or water, as control) in a final volume of 2 µL. The qPCR data were normalized on U6 RNA levels, and the fold change was calculated by monitoring doRNA (B) or C-doRNA (C) versus the control sample (mean ± SEM; n = 4 independent experiments). The symbol (+) denotes the presence, and (−) the absence of the indicated component. The horizontal dotted red line indicates a fold change of 1 (no change). **** p < 0.0001 (one-way ANOVA with Holm–Sidak’s post hoc test).

Figure 6

doRNA and C-doRNA detection by splinted 5′ adapter RT-qPCR is linear over a range of 7 logs. (A,B) Standard curves were established using synthetic doRNA (A) or C-doRNA (B) serial dilution to calculate doRNA or C-doRNA copy numbers in various cell and tissue samples.

Figure 7

Splinted 5′ ligation RT-qPCR validation of small RNA-Seq data. Correlation between the relative abundance of doRNA (A), C-doRNA (B), miR-25 (C) and miR-30a (D) assessed by splinted 5′ ligation RT-qPCR or small RNA-Seq. qPCR data are represented as the average copy number from 3 technical replicates, while small RNA-Seq data are represented as the number of reads per million for one replicate. Linear models were used to plot the line of best fit for each correlation with confidence of 95%; the R² value of the correlation is indicated.