Comparative Analysis of rRNA Removal Methods for RNA-Seq Differential Expression in Halophilic Archaea

Abstract

Despite intense recent research interest in archaea, the scientific community has experienced a bottleneck in the study of genome-scale gene expression experiments by RNA-seq due to the lack of commercial and specifically designed rRNA depletion kits. The high rRNA:mRNA ratio (80-90%: ~10%) in prokaryotes hampers global transcriptomic analysis. Insufficient ribodepletion results in low sequence coverage of mRNA, and therefore, requires a substantially higher number of replicate samples and/or sequencing reads to achieve statistically reliable conclusions regarding the significance of differential gene expression between case and control samples. Here, we show that after the discontinuation of the previous version of RiboZero (Illumina, San Diego, CA, USA) that was useful in partially or completely depleting rRNA from archaea, archaeal transcriptomics studies have experienced a slowdown. To overcome this limitation, here, we analyze the efficiency for four different hybridization-based kits from three different commercial suppliers, each with two sets of sequence-specific probes to remove rRNA from four different species of halophilic archaea. We conclude that the key for transcriptomic success with the currently available tools is the probe-specificity for the rRNA sequence hybridization. With this paper, we provide insights into the archaeal community for selecting certain reagents and strategies over others depending on the archaeal species of interest. These methods yield improved RNA-seq sensitivity and enhanced detection of low abundance transcripts.

Keywords: RNAs-seq; archaea; halophiles; rRNA removal; transcriptomics.

Figure 1

Slowdown in Archaeal RNA-Seq publications in recent years. Lines depicting number of publications per year detected in the NCBI PubMed databased searched with the terms “Archaea” (red), “RNA-Seq” (blue), and “Archaea RNA-Seq” (green), plotted on log-scale y-axis. Dotted line at 2018 marks discontinuation of the Illumina RiboZero kit.

Figure 2

Percentage of rRNA remaining in halophile RNA by using the discontinued Ribozero kit (RZ). Each dot denotes one sample, with light orange dots representing Hbt. salinarum and blue dots representing Hfx. volcanii samples. Horizontal bars represent the median percentage of rRNA remaining.

Figure 3

rRNA removal using alternative methods in Hbt salinarum. Each dot represents the percentage of counts mapping to rRNA genes after using no removal (brown), New Ribozero kit (RZ+, dark orange), NEBNext kit with bacterial probes (NEB-B, orange), and NEBNext kit with HVO probes (NEB-HVO, peach). Horizontal bars represent the median value.

Figure 4

Increasing RNAse digestion time is less important than probe sequence identity for efficient rRNA removal. (A) Dotplot showing percentage of counts mapping to rRNA genes after using the NEB-HVO method on HBT total RNA samples after 30 min (brown) or minutes 120 (light orange) of RNAseH digestion. “NEB30 (2)” samples to the right of the dotted line were processed and sequenced in a different batch. Horizontal bars represent the median value. (B) Percentage of custom-designed HVO probes classified into 16S (black) and 23S (grey). Levels of sequence identity of HVO probes with Hbt. salinarum (HBT) 16S and 23S rRNA genes are shown on the X-axis, whereas the percentage of total probes at each sequence identity level is shown on the Y-axis. (C) Percentage of total reads mapping to either 16S (left panel) or 23S rRNA (right panel) genes of HBT using three different rRNA removal method—none (brown), NEB-B (dark orange), and NEB-HVO (light orange).

Figure 5

Species-specific probes efficiently remove rRNA from target species. Dotplots showing the percentage of rRNA remaining after using probes with sequences specific for Hfx. volcanii (HVO) rRNA. Dark blue dots represent %rRNA remaining in individual replicate samples depleted with NEBNext Core Reagent Set (“NEB-HVO” method). Light blue dots represent %rRNA remaining in individual replicate samples depleted with the siTools RiboPool kit (“rP-HVO”). Horizontal bars represent the median value.

Figure 6

Panarchaea (rP-PA) kit efficiently removes rRNA from total RNA across halophilic species. Dotplots showing the percentage of the remaining counts mapping to rRNA genes in Hbt. salinarum (HBT, orange), Hfx. volcanii (HVO, blue), Hfx. mediterranei (HVO, purple), and Hca. hispanica (HCA, grey). Horizontal bars represent the median value of three biological replicate samples.

Figure 7

Choice of removal method does not affect relative abundance of mRNAs. Correlations between relative abundance of each gene after different rRNA removal methods in (A) Hbt. salinarum (HBT) and (B) Hfx. volcanii (HVO). Each dot represents the percent of total normalized reads for each gene (see Methods section). Methods shown here are “Control” (no removal), “RZ” (using discontinued RiboZero kit), “NEB-HVO” (using NEBNext kit with custom HVO probes), “NEB-120” (NEBNext kit with custom HVO probes and 120 min of RNAse digestion), “rP-PA” (siTools riboPOOL method using Panarchaeal probes), and “rP-HVO” (siTools riboPOOL method using HVO-specific probes).

Figure 8

More complete rRNA removal leads to increased detection of lowly expressed genes. Number of genes with zero (open circles) and <5 (filled circles) reads detected in sequencing samples treated with different rRNA removal methods in Hbt. salinarum. The darker the circle color, the more complete the rRNA removal for each method: riboPOOL Panarchaea (rP-PA, black); NEBNExt with HVO-specific probes (NEB-HVO, dark grey); NEBNext with bacterial probes (NEB-B, grey); no removal (none, light grey).