Phased nucleotide inserts for sequencing low-diversity RNA samples from in vitro selection experiments

Abstract

In vitro selection combined with high-throughput sequencing is a powerful experimental approach with broad application in the engineering and characterization of RNA molecules. Diverse pools of starting sequences used for selection are often flanked by fixed sequences used as primer binding sites. These low diversity regions often lead to data loss from complications with Illumina image processing algorithms. A common method to alleviate this problem is the addition of fragmented bacteriophage PhiX genome, which improves sequence quality but sacrifices a portion of usable sequencing reads. An alternative approach is to insert nucleotides of variable length and composition ("phased inserts") at the beginning of each molecule when adding sequencing adaptors. This approach preserves read depth but reduces the length of each read. Here, we test the ability of phased inserts to replace PhiX in a low-diversity sample generated for a high-throughput sequencing based ribozyme activity screen. We designed a pool of 4096 RNA sequence variants of the self-cleaving twister ribozyme from Oryza sativa For each unique sequence, we determined the fraction of ribozyme cleaved during in vitro transcription via deep sequencing on an Illumina MiSeq. We found that libraries with the phased inserts produced high-quality sequence data without the addition of PhiX. We found good agreement between previously published data on twister ribozyme variants and our data produced with phased inserts even when PhiX was omitted. We conclude that phased inserts can be implemented following in vitro selection experiments to reduce or eliminate the use of PhiX and maximize read depth.

Keywords: in vitro selection; low-diversity; phased inserts; ribozymes; sequencing.

FIGURE 1.

Twister ribozyme library design and in vitro protocol. (A) Secondary and tertiary structure of the twister Osa-1-4 ribozyme (Liu et al. 2014; Rose and Hildebrand 2015). The library contained six randomized nucleotide positions indicated by the red nucleotides. The triangle in the secondary structure indicates the cleavage site, and the black nucleotides are the cleaved product. (B) Illustration of the protocol for cotranscriptional self-cleavage and phased nucleotide insertion during template switching reverse transcription.The DNA library is ordered as the template strand for transcription, with the T7 promoter at the 3′-end, and a primer binding sequence at the 5′-end (linker). Active variants self-cleave during transcription by T7 RNA polymerase. Cleaved and uncleaved RNA products are reverse-transcribed with template switching using the linker sequence for primer binding, and a pool of four phased template switching oligonucleotides. These phased inserts are incorporated into the the cDNA transcripts during reverse transcription. The resulting single stranded cDNA products with phased inserts are amplified with index primers to add full adaptors for high-throughput sequencing (Illumina).

FIGURE 2.

Calculation of positional entropy from simulated sequencing data. Entropy was calculated for four simulated twister ribozyme library samples based on the balance of nucleotides that would be observed at each sequencing cycle. At each position in the read (y-axis), the maximum entropy = 2 occurs when there is an equal probability of all four nucleotides. The minimum entropy = 0 when the same nucleotide occurs at the same position in all sequences. Positional entropy is shown for simulated sequencing runs representing the four conditions that were experimentally sequenced. The control library without phased nucleotide inserts or PhiX (light blue), only 25% PhiX (light red), addition of only phased inserts (dark blue), or both phased inserts and PhiX (dark red).

FIGURE 3.

Quality scores of sequencing runs. (A) Mean sequencing quality scores per position in the read for samples without phased nucleotide inserts or PhiX (light blue), only 25% PhiX (light red), addition of only phased nucleotide insertions (dark blue), or both phased insertions and PhiX (dark red). (B) Distribution of mean sequence quality (Phred score) for sequencing runs. (C) Distribution of minimum sequence quality (Phred score) for sequencing runs.

FIGURE 4.

Comparison of ribozyme activities with published data. Heatmap visualization of the activity (fraction cleaved) of the twister ribozyme mutants with previously published values (Kobori and Yokobayashi 2016). Nucleotide identities of mutations are shown as row and column labels. Double mutants are depicted at the intersection of two mutations. The diagonal contains compensatory double mutations that maintain Watson–Crick base pairs. Corresponding heat maps are also shown for the four sequencing runs presented in this study. Correlations between our data and the published values are shown below each heatmap. Pearson correlation coefficients (r) are shown below each correlation.