Next-generation sequencing technologies have revolutionized the study of small RNAs (sRNAs) on a genome-wide scale. However, classical sRNA library preparation methods introduce serious bias, mainly during adapter ligation steps. Several types of sRNA including plant microRNAs (miRNA), piwi-interacting RNAs (piRNA) in insects, nematodes and mammals, and small interfering RNAs (siRNA) in insects and plants contain a 2′-O-methyl (2′-OMe) modification at their 3′ terminal nucleotide. This inhibits 3′ adapter ligation and makes library preparation particularly challenging. To reduce bias, the NEBNext kit (New England Biolabs) uses polyethylene glycol (PEG), the NEXTflex V2 kit (BIOO Scientific) uses both randomised adapters and PEG, and the novel SMARTer (Clontech) and CATS (Diagenode) kits avoid ligation altogether. Here researchers from the CNRS compared these methods with Illumina’s classical TruSeq protocol regarding the detection of normal and 2′ OMe RNAs. In addition, they modified the TruSeq and NEXTflex protocols to identify conditions that improve performance.
Among the five kits tested with their respective standard protocols, the SMARTer and CATS kits had the lowest levels of bias but also had a strong formation of side products, and as a result performed relatively poorly with biological samples; NEXTflex detected the largest numbers of different miRNAs. The use of a novel type of randomised adapters called MidRand-Like (MRL) adapters and PEG improved the detection of 2′ OMe RNAs both in the TruSeq as well as in the NEXTflex protocol.
a Sequence bias of the various protocols. The standard deviation of the proportion of reads corresponding to each of the unmodified RNAs 1–6 was taken as a measure of sequence bias. Shown are the data for each replica of the different protocols. We did not consider variation among the 2’ OMe RNAs here, as additional variability is introduced by 2’ OMe bias. The type of adapters used for the various protocols and the presence (“+”) or absence (“-”) of PEG and a purification step after 3’ adapter ligation is indicated. Also the presence or absence of overnight ligation at 16 °C is indicated, with “absence (-)” meaning standard ligation at 28 °C (TS protocols) or 22 °C (Nf protocols). b 2’ OMe bias of the various protocols. The ratios of the total numbers of reads for the unmodified RNAs (RNA1–6) and for the 2’ OMe RNAs (RNA-OMe1–6) were determined for each protocol and in each separate experiment. Shown are the mean values of at least two independent experiments and the standard deviations are indicated by error bars. c Percentage of the total numbers of reads corresponding to side products. The percentages of raw reads with inserts < 10 nt (considered adapter dimers and eliminated after trimming) are indicated in green. Blue bars represent inserts ≥10 nt, < 25 and ≠ 21 nt that did not correspond to RNA(OMe)1–6. Yellow bars represent inserts ≥25 nt that did not correspond to RNA(OMe)1–6. Shown are the mean values of at least two independent experiments and the standard deviations are indicated by error bars
While it is commonly accepted that biases in sRNA library preparation protocols are mainly due to adapter ligation steps, the ligation-free protocols were not the best performing methods. The modified versions of the TruSeq and NEXTflex protocols provide an improved tool for the study of 2′ OMe RNAs.