Lasy-Seq: a high-throughput library preparation method for RNA-Seq and its application in the analysis of plant responses to fluctuating temperatures

Abstract

RNA-Seq is a whole-transcriptome analysis method used to research biological mechanisms and functions but its use in large-scale experiments is limited by its high cost and labour requirements. In this study, we have established a high-throughput and cost-effective RNA-Seq library preparation method that does not require mRNA enrichment. The method adds unique index sequences to samples during reverse transcription (RT) that is conducted at a higher temperature (≥62 °C) to suppress RT of A-rich sequences in rRNA, and then pools all samples into a single tube. Both single-read and paired-end sequencing of libraries is enabled. We found that the pooled RT products contained large amounts of RNA, mainly rRNA, causing over-estimations of the quantity of DNA and unstable tagmentation results. Degradation of RNA before tagmentation was found to be necessary for the stable preparation of libraries. We named this protocol low-cost and easy RNA-Seq (Lasy-Seq) and used it to investigate temperature responses in Arabidopsis thaliana. We analysed how sub-ambient temperatures (10-30 °C) affected the plant transcriptomes using time-courses of RNA-Seq from plants grown in randomly fluctuating temperature conditions. Our results suggest that there are diverse mechanisms behind plant temperature responses at different time scales.

Figure 1

Comparison of the RNA-Seq library preparation methods. Steps modified in Lasy-Seq are shown on the right with red characters. In the conventional method (left), the high-throughput RNA-Seq required parallel preparation of all individual samples throughout all experimental steps. In Lasy-Seq, enrichment of mRNA was not required, and all samples were pooled into a single tube after the RT step, by adding unique index sequencing to each sample at the RT step. SI and LI indicate small-input and large-input total RNA, respectively. SR and PE indicate single-read sequencing and paired-end sequencing, respectively.

Figure 2

RT at high temperature and RNase treatment were important for stable library preparation in Lasy-Seq. (A) Comparison of the distribution of the reads mapped on 25S-5.8 S rRNA reverse transcribed at 50 °C and 62 °C. RT of non-poly-A tailed RNAs were observed from internal A-rich regions. RT at higher temperature suppressed the RT from internal A-rich regions of non-poly-A tailed RNA. (B) List of the delta-Cp values in RT-qPCR on genes with and without poly-A tails. (C) The amount of RNA in reaction solutions after RT. If mRNA occupied 4% of the total amount of RNA in a cell, the rate RNA remained after RT of total RNA became 25 times larger than cDNA. (D) Effect of RNA addition on the measurement of DNA concentrations. Concentrations of DNA were determined for samples with constant amounts of DNA (0.7 ng) and different amounts of RNA (from 0 to 30 times larger than the DNA quantity). DNA concentration was over-estimated in RNA-added samples which contained RNA concentrations more than 10 times larger those of DNA. In the table, “DNA conc.” and “RNA conc.” indicate true concentration of measured liquids. “Measured value of DNA conc.” means the concentration determined by QuantiFluor dsDNA System and Quantus Fluorometer. (E) The plotted concentrations and DNA:RNA rates.

Figure 3

Evaluation of the false-assignment rates associated with sample pooling at early stages. (A) Flow of the library preparation for evaluation of false-assignment rates among samples. Early-pooled sets were pooled before the tagmentation step, while late-pooled sets were individually prepared until purification after PCR. All samples were pooled prior to sequencing. (B) Number of ERCC reads detected in each sample. Numbers shown above the bar-plot indicate the read number of ERCC. Conditions of the experiment for each sample were shown by the colours of the bar and indicated over the bar-plot.

Figure 4

Comparison of quantitative performance of a conventional method and Lasy-Seq. (A) Plot of log₂(rpm + 1) of all genes in biological replicates of Lasy-Seq library. (B) Plot of log₂ (rpm + 1) of all genes on the nuclear genome of a RNA-Seq library of the conventional method and of Lasy-Seq. (C) The read depth distribution of the conventional method and Lasy-Seq. X-axis indicates the position from the 3′ end of each transcript. Y-axis indicates the sum of the depth of the mapped read onto all genes on the nuclear genome in six libraries of 5 M total reads. (D) Number of detected genes in the conventional method and Lasy-Seq library with one, two, three, four and five million of the subsampled reads. (E) The number of DEGs between light and dark conditions detected with each RNA-Seq library preparation method (FDR = 0.05).

Figure 5

Temperature settings in the temperature response experiment. (A) The three sets of temperature conditions. Plants were grown at 20 °C for 8 days and then at changing temperature conditions for 3 days. Sampling was conducted from 14 to 21 day after sowing (d.a.s.), indicated by red characters. (B) Diagram of the temperatures of the three sets from 8 d.a.s to 21 d.a.s. (C) Correlation of the temperature between sampling day and the days prior to sampling. Horizontal axis shows temperature (°C) on the sampling day and vertical axis indicates the temperatures 1,2 and 3 days prior to sampling (from left to right, respectively). The “Adjusted-p” indicated adjusted p-value (FDR) and “r” indicated Pearson’s correlation coefficients.

Figure 6

The correlation between the transcriptome and the temperature. (A) Flow of the analysis of the correlation between gene expressions and temperatures. (B) Distribution of the correlation coefficients for each gene between gene expression levels and the temperature of the sampling day and 1, 2 and 3 day prior to sampling (from top panel to bottom panel, respectively). Each circle indicates each gene. Red and orange circles indicate genes for which significant relationships between the expression and the temperature (adjusted p-value < 0.1) were detected. Red circles represent genes with correlation coefficients of more than 0.05.

Figure 7

Genes correlated to temperature on sampling day. (A) Expression of PCL1 and GI genes. The horizontal axis indicates temperature settings for each sample on sampling day. The vertical axis indicates expression of each gene by log₁₀ (rpm + 1). Each circle indicates each sample (n = 45) and the red lines are regression lines. “Cor.coef” indicates correlation coefficients. (B) Schematic diagram of the changes in amplitudes of the circadian oscillations of GI correlated to temperature changes reported in a previous study. The lines with “high”, “middle” and “low” represent the circadian oscillations of GI under each temperature condition. A green broken line indicates sampling times in the present study and expression of GI at the time became smaller at higher temperatures. (C) Expression of LFY and the regulator or target genes of LFY. Horizontal axis and vertical axis are same as (A).

Figure 8

Genes that responded to temperatures one day prior to sampling. Expression levels of CBL6 (top panel), FPGS1 (middle panel), and NUC2 (bottom panel) were plotted. The horizontal axis indicates temperature settings for each sample on each sampling day (left three panels) and one day prior to sampling (right three panels). The vertical axis means show expression for each gene by log₁₀ (rpm + 1). Each circle indicates each sample (n = 45) and red lines are regression lines (drawn only in case of adjusted p-value < 0.1).