The advent of functional genomics and the availability of next generation sequencing (NGS) technologies have dramatically changed the approach to studying gene expression. Massive RNA sequencing provides a detailed snapshot of the total RNA (the “transcriptome”) present at a given time in a cell. The transcriptome comprises coding (mRNA) and non-coding RNA (rRNA, tRNA, regulatory RNA and other RNA species). Quantitative differences in the gene expression patterns, either between cells grown in different conditions or between a mutant and its parental strain, can be identified in their entirety through this approach. However, in order to obtain reliable information, it is crucial that RNA sequencing achieves a sufficient “coverage”, enough to detect even rare RNA species. In bacterial cells, a disadvantage of the transcriptomic approach is the high amount of ribosomal RNA (rRNAs), which accounts for more than 95% of total cellular RNA, greatly reducing useful transcript coverage. Thus, efficient removal of rRNA is critical for successful transcriptome profiling. Unlike for eukaryotic mRNA, polyadenylation of bacterial mRNAs is limited and is mostly involved in targeting mRNA for degradation by PNPase ; hence, bacterial mRNA cannot be readily isolated from other RNA sources by hybridization to immobilized poly-T or enriched through reverse transcription with poly-T primers. Therefore, a major challenge in RNA-seq applications in bacterial cells is the enrichment for all transcript species other than rRNA and tRNA.
Papers describing the use of high-throughput sequencing for transcriptomics in bacteria have used mRNA enrichment methods usually based on depletion of rRNA and other RNAs [2–4], utilizing two alternative approaches: (i) hybridization capture of rRNAs by antisense oligonucleotides followed by pull down through binding to magnetic beads, (ii) degradation of processed RNA such as mature rRNA and tRNA by a 5′–3′ exonuclease that specifically digests RNA species with a 5′-monophosphate end.
The rRNA capture approach using the MICROBExpress Kit (Ambion) has widely been applied in RNA-seq studies [3–8], including metatranscriptomics [9, 10]. Usage of the RiboMinus Bacteria Transcriptome Isolation Kit (Invitrogen), based on the same method, has only been reported in one study so far . The rRNA capture approach is the only method suitable for precise quantitative analysis. However, as it is based on 16S and 23S rRNA specific capture probes, depletion efficiency of these kits varies between bacterial species. An alternative approach implemented a subtractive hybridization protocol using probes targeting bacterial, archaeal and eukaryotic fractions of environmental rRNA pools . Finally, a recent work  reports a very extensive comparative analysis of five different rRNA removal methods, emphasizing the efficiency of the Ribo-Zero rRNA removal kit from Epicentre. The Ribo-Zero kit, also based on rRNA capture, proved to be very efficient both on pure cultures and on faecal samples, while preserving mRNAs relative abundance.
Alternatively to the capture-based methods, rRNA removal can also be achieved through its degradation by specific enzymes. An example is the mRNA-ONLY Prokaryotic mRNA Isolation Kit (Epicentre Biotechnologies), based on selective degradation of processed RNAs by the enzyme terminator 5′-phosphate-dependent exonuclease (TEX). This enzyme exclusively degrades RNA molecules carrying a 5′-monophosphate, i.e., processed RNA such as rRNA and tRNA, while mRNAs, carrying a 5′-triphosphate group, are not affected . This method can be useful for the analysis of very complex samples by RNA-seq (e.g., environmental metatranscriptomics) , but it only provides semi-quantitative evaluation of gene expression levels. In some instances, selective rRNA methods have been used in combination with subtractive hybridization to optimize rRNA removal [15, 16]. An additional advantage of rRNA degradation with TEX consists in the enrichment of primary transcripts with 5′-triphosphate ends, thus allowing identification of transcription start sites [14, 17, 18]. This methodology, termed differential RNA-seq (dRNA-seq), is extremely informative for promoter mapping and identification of small RNAs. Finally, another enzymatic method for mRNA enrichment that makes use of duplex-specific nuclease (DSN) to remove rRNA has been applied with good results, both in terms of mRNA coverage and robustness of mRNAs relative abundances, in transcriptome profiling of Escherichia coli grown in four different conditions .
In this work we tested the Ovation Prokaryotic RNA-Seq System kit for bacteria ribosomal RNA removal developed by NuGEN (NuGEN Technologies, San Carlos, CA, USA). Unlike the approaches described so far, which are based on rRNA removal or degradation, the Ovation kit relies on the synthesis of first and second strand cDNA using a random primer mix selectively designed to enrich the mRNA portion of bacterial total RNA. The selective random primers are designed against a sequence database composed of 50 bacterial and archaeal strains representing all of the major phylogenetic subgroups. The predicted binding site density of these primers on target (mRNA) and non-target (rRNA) transcripts is nearly identical across these species. The resulting cDNA is compatible with NuGEN’s Encore™ NGS Library Systems as well as other library workflows using double-stranded cDNA as input for the creation of sequencing libraries.
This new method was tested either in the absence of further treatments or in combination with an rRNA capture-based approach, i.e., the MICROBExpress Kit from Ambion. A comparison of the results obtained from rRNA removal procedures based on different chemistries (capture with probes versus retro-transcription with selective random primers) and from libraries prepared with different protocols (Illumina TruSeq RNA libraries versus NuGEN’s Encore™ NGS libraries) was performed in order to evaluate the efficiency of the two methods, either alone or in combination, and to test the robustness of the protocols. The rRNA removal efficiency of the two kits, either separately or in combination was evaluated on RNA extracted from Burkholderia thailandensis in two different growth conditions. This bacterium was chosen because of its importance as a model organism for pathogenic species of Burkholderia such as B. pseudomallei, the etiological agent of melioidosis , and since its genome is characterized by a high GC content.
One of the main goals of this work was the reduction of sequencing costs: to this aim, we evaluated whether the two rRNA removal treatments, tested either separately or in combination, could result in a scaling down of the sequencing size (total reads produced) while preserving the whole transcriptome coverage. Our results show that the combination of the two kits leads to optimal results in terms of rRNA removal, without introducing a significant bias on relative mRNA abundances, and allow the sequencing of a GC rich bacteria transcriptome with less than 10 millions reads.