Background

Although small RNAs, which are 20-30 nt long regulatory non-coding RNA molecules, constitute a tiny percentage of the total cellular RNA, they have a huge impact on many cellular processes (including, gene regulation, organism development, adaptation to changing environment and so on) (Jost et al., 2011). There are several classes of small RNA (Youngman and Claycomb, 2014) with microRNAs (miRNAs) being the most extensively studied (Bartel, 2009).

Besides being of a great interest for basic research, miRNAs hold promise as diagnostic markers. Due to their exceptional stability, intact miRNAs are present in preserved tumor samples (formalin-fixed paraffin-embedded [FFPE]), which exhibit a near total mRNA degradation (Nagy et al., 2016). Moreover, stable miRNAs circulate in body fluids such as plasma, urine and sputum of healthy and diseased patients (Mitchell et al., 2008). Liquid biopsy of circulating miRNAs can serve as potential markers for a broad range of human diseases especially for diseases for which ordinary biopsies cannot be obtained, such as cardiovascular diseases, and psychiatric diseases (Kim, 2015; Backes et al., 2016).

While assessing the repertoire and abundance of cellular miRNAs became an important aspect in many studies, the tools to do so lag behind. High-throughput sequencing (HTS) is a gold standard method for mRNA profiling, extensively used in both basic research and clinical settings. In striking contrast, HTS profiling of miRNAs is not widespread, as preparation of small RNA libraries remains technically challenging.

To become compatible with Illumina or other HTS platforms, small RNAs are ligated from both sides (5’ and 3’) to oligonucleotide adapters and then reverse transcribed. The resulting small RNA library is then amplified by PCR with primers complementary to the 5’ and 3’ adapters that include sequences recognized by the sequencing machinery (Lu et al., 2007). If small RNAs are not separated from other RNA species, in particular from tRNA, which are highly abundant and close in size (~70 bp), the resulting libraries will be highly contaminated by irrelevant sequences. In addition, 3’ and 5’-adapters ligate to each other, forming adapter-dimers, which contaminate libraries with null sequences. Thus, for generation of a high-quality sRNA library, a preparation protocol must enable separation of small RNA from total RNA and separation of the ligation products from free adapters. Size-selection on Solid Phase Reversible Immobilization (SPRI) paramagnetic beads (Lundin et al., 2010), used in preparation of mRNA and DNA libraries, was not suitable for preparation of small RNA libraries, as the lower limit of size discrimination is 100 nt (Paithankar et al., 1991). The only remaining option, electrophoresis on PAGE gel, is cumbersome, time-consuming, requires technical skills, results in a huge loss of the product and eliminates the possibility of automatization. Preventing adapter-dimer contamination by molecular biology strategies (Kawano et al., 2010; Vigneault et al., 2012) is not a good alternative to size-selection as it does not solve the problem of contamination by irrelevant sequences.

To allow easy size selection of short nucleic acid molecules, we modified the SPRI method by using a combination of two crowding agents, Polyethylene Glycol (PEG) and Isopropanol. This modification enables separation of nucleic acid fragments shorter than 100 nt, differing in length by only 20 nt (Fishman et al., 2018). Exploiting this new separation method, we designed a protocol for preparation of small RNA libraries for high-throughput sequencing based on Lau et al. (2001) and we named the protocol QsRNA-seq. The input material for the protocol is either total RNA or low molecular weight RNA fraction (LMW RNA, < 200 nt). During the protocol three SPRI-based size selection steps are performed, each using different conditions, to allow separation of sRNA from total RNA and of ligation products from free adapters after each ligation (for the protocol outline see the Procedure section below). In addition, the protocol supports the possibility of multiplexing samples by adding a 4-nt barcode to the 5’-adapter. The protocol also supports using unique molecular identifiers (UMI) for reduction of ligation bias (Fuchs et al., 2015) and correction of amplification bias (Kivioja et al., 2011) by adding an 8-nt random sequence to the 5’-adapter as well. To correct the amplification bias, identical small RNA sequences with the same UMI are considered to be amplification products and merged to one sequence (i.e., collapsed) during data processing. By comparing libraries generated from the same RNA using adapters with the same barcode either with UMI or without UMI, we found that adding UMI significantly reduced both biases (Fishman et al., 2018).

QsRNA-seq has several significant advantages over the existing protocols. The entire protocol can be performed in one day and does not require high manual skills. Because it does not require any gel use there is less material lost, which enables small quantities of the input RNA to be analyzed. Moreover, it allows simultaneous processing of many samples as well as automation, turning the preparation of small RNA libraries into a routine procedure. In addition, usage of UMI allows reduction of library preparation biases, resulting in more reliable quantitation. To summarize, QsRNA-seq is an easy-to-use protocol that generates reliable and quantifiable high-throughput small RNA libraries.