Improved precision, sensitivity, and adaptability of Ordered Two-Template Relay cDNA library preparation for RNA sequencing

Abstract

Sequencing RNAs that are biologically processed or degraded to less than ~100 nucleotides typically involves multi-step, low-yield protocols with bias and information loss inherent to ligation and/or polynucleotide tailing. We recently introduced Ordered Two-Template Relay (OTTR), a method that captures obligatorily end-to-end sequences of input molecules and, in the same reverse transcription step, also appends 5' and 3' sequencing adapters of choice. OTTR has been thoroughly benchmarked for optimal production of microRNA, tRNA and tRNA fragments, and ribosome-protected mRNA footprint libraries. Here we sought to characterize, quantify, and ameliorate any remaining bias or imprecision in the end-to-end capture of RNA sequences. We introduce new metrics for the evaluation of sequence capture and use them to optimize reaction buffers, reverse transcriptase sequence, adapter oligonucleotides, and overall workflow. Modifications of the reverse transcriptase and adapter oligonucleotides increased the 3' and 5' end-precision of sequence capture and minimized overall library bias. Improvements in recombinant expression and purification of the truncated Bombyx mori R2 reverse transcriptase used in OTTR reduced non-productive sequencing reads by minimizing bacterial nucleic acids that compete with low-input RNA molecules for cDNA synthesis, such that with miRNA input of 3 picograms (less than 1 fmol), fewer than 10% of sequencing reads are bacterial nucleic acid contaminants. We also introduce a rapid, automation-compatible OTTR protocol that enables gel-free, length-agnostic enrichment of cDNA duplexes from unwanted adapter-only side products. Overall, this work informs considerations for unbiased end-to-end capture and annotation of RNAs independent of their sequence, structure, or post-transcriptional modifications.

Keywords: OTTR; non-coding RNA; non-templated nucleotide addition; reverse transcriptase; template jumping; terminal transferase.

Figure 1:

Improvements in the OTTR library synthesis workflow. Previous OTTR conditions are in black or gray text, and improvements are given in red text and mark-out of black text with a red line. The left side explains the steps, while the right side gives buffer recipes.

Figure 2:

Evaluating imprecision of end-to-end sequence capture at RNA 3′ ends. A, Mechanisms, numbered by Roman numeral, of 3′-end capture in OTTR. Column one illustrates five specificities for capturing an input-template 3′ end, following the color legend from Figure 1. Phosphodiester bonds are represented by circles: gray for chemically synthesized bonds, concentric double circle for BoMoC-polymerized, unfilled for input template RNA. Strand 3′ end symbols are a small black circle for dideoxynucleotide, a small open circle for 3′-OH, and a triangle for 3′-OH replacement with an unextendible carbon linker. The red octagon indicates the presence of a 5′ fluorescent dye. Column two specifies the labeling status of the input-template 3′ end. Column three identifies the primer-duplex 3′ overhang supporting a template jump. Column four details the impact on 3′ sequence coverage. Column five categorizes the 3′-end capture mechanism. B, Library-wide CV of miRXplore miRNA counts and mean fraction of alignments with complete 3′-end coverage (3′ precision). BoMoC, yPAP (for 30 or 120 minutes), or no enzyme (Unlabeled) was used for 3′ labeling. C-E, Panel pairs depict 3′-end coverage with BoMoC (left) or no enzyme (right) for 3′ labeling. The miRNA sequence and name, and relative rank-order, are indicated at top. At bottom, position “0” is the miRNA 3′ end. Right panels have at bottom the miRNA sequence color-coded by alignment inclusion (pink) or exclusion (black). The Read2 (R2) primer sequence and adjacent +1 T overhang (+T) are in blue, and reference to the capture mechanism detailed in (A) is indicated. Input RNA 3′ sequence excluded by adapter trimming is indicated in black. The miRNA mmu-miR-33–5p is miRBase: MIMAT0000667, the miRNA hsa-miR-599 is miRBase: MIMAT0003267, and the miRNA mghv-miR-M1–3-3p is miRBase: MIMAT0001566.

Figure 3:

Optimization of 3′ labeling via BoMoC sequence and reaction buffers. A, Library-wide CV and 3′ precision of miRXplore miRNA as in Figure 2B, using BoMoC WT or variant enzymes. Replicate cDNA libraries have the same color. B, Comparison of 3′ precision under different BoMoC working-stock enzyme storage buffer conditions. Significant differences in 3′ precision (denoted by asterisks) were determined by a paired, two-sided student’s t-test for each miRNA, with p-values adjusted by Bonferroni correction. Each library was benchmarked against a library prepared with enzyme diluted in 15 mM DTT in pH 6.0 diluent buffer. C, Comparison of 3′ precision using different BoMoC proteins, incubation times, and ddNTP(s) for 3′ labeling. Significant differences were measured as described in (B).

Figure 4:

Evaluation of imprecision of end-to-end sequence capture at RNA 5′ ends. A, Mechanisms, numbered by Roman numeral, of precise and imprecise template-jumping to the 3′rC adapter template. Column one illustrates two possible and one unlikely specificity of sequence-junction formation following the color legend of Figures 1 and 2A. Phosphodiester bonds are represented by circles: gray for chemically synthesized bonds, concentric double circle for BoMoC-polymerized, unfilled for input template RNA. Strand 3′ end symbols are a small black circle for dideoxynucleotide, a small open circle for 3′-OH, and a triangle for 3′-OH replacement with an unextendible carbon linker. The red octagon indicates the presence of a 5′ fluorescent dye. Column two defines the NTA that would prime 3′rC adapter-template capture. Column three defines whether the mapped template 5′-end was precise. B,C, Comparison of mean fraction of alignments with 5′ precision for libraries prepared with different 3′rC adapter template designs (Table 5). Significant differences in 5′ precision (denoted by asterisk) were determined by a paired, two-sided student’s t-test for each miRNA, with p-values adjusted by Bonferroni correction. See complete summary statistic in Table 4.

Figure 5:

OTTR performance across different amounts of input RNA. A, Representative replicates of OTTR cDNA resolved by 8% dPAGE and detected using the 5′ Cy5 or IRD800 primer fluorophore, with miRXplore miRNA input in pg and 5′ primer fluorophore denoted above. cDNA product size-selection range is denoted by the left open bracket (black). On the right, adapter-dimer and miRXplore OTTR cDNA products are indicated with schematics using the color key from Figure 1. B, Correlation matrix of miRNA read counts from cDNA libraries produced using a titration from 4 to 500 pg total miRXplore RNA. Replicate Cy5 and IRD800 libraries for each miRNA input amount were averaged by their counts per million reads (CPM) and compared by Spearman’s correlation coefficient (ρ). miRXplore miRNA input in pg, 5′ primer fluorophore, and library-wide CV are denoted. C, Composition bar plots of the fraction of the total library, excluding reads 17 nucleotides or shorter, that mapped to miRXplore miRNA, OTTR adapter sequences, BoMoC expression plasmid, or E. coli genome, or were unmapped. Individual replicate libraries are presented in pairs.

Figure 6:

Improvements in BoMoC protein purification. A, Composition bar plots of mapped reads, excluding reads 17 nucleotides or shorter and adapter-mapping reads, for miRXplore cDNA libraries made using 20 pg input RNA and different protein purifications (preps). Variations to protein purification (Supplemental Fig. 4) are described using filled circles to indicate the presence of a variable E. coli strain or purification step in each prep. Prep 1 and 2 were BoMoC W403AF753A and were purified after splitting the same cell lysate in half. B, Subcategories of read mapping for the E. coli nucleic acid category in (A). C, Read length distribution plots of for the subcategories of read mappings in (B). Note that the y-axis for each plot has a different scale. D,E, Composition bar plots as described for (A-B) for cDNA libraries from 20 pg miRXplore input. In addition to 20 pg miRXplore RNA input, a 500 pg miRXplore RNA input cDNA library was produced; the pair were used to extrapolate minimum input that would recover 9:1 miRNA:E. coli and expression-plasmid reads, with that amount indicated at top. Preps 3 and 8 were BoMoC F753A; preps 4, 5, 6, and 7 were BoMoC W403AF753A; and prep 9 was BoMoC WT. F, Titration results for data from Figure 5 and additional results for cDNA libraries prepared using preps 8 and 9 for 3′ labeling and cDNA synthesis, respectively. The y-axis shows the read count ratio of miRXplore:contaminants (i.e., E. coli and expression plasmid sequences). Both axes are on a log₁₀ scale. G, Composition bar plots as described for (A), here for libraries constructed with 0.2, 1, 4, 20, and 500 pg of miRXplore RNA and using preps 8 and 9 for 3′ labeling and cDNA synthesis, respectively. One representative replicate from the data in (F) was used for the bar plot. Red horizontal dashed lines (10% and 90%) are included as visual aids.

Figure 7:

Gel-free OTTR by capture of biotinylated-input OTTR cDNA duplexes. A, The 3′ labeling activity of W403AF753A BoMoC with various d(d)RTPs resolved by 12% dPAGE and detected by the 5′ IRD800 fluorophore of the 3′NN single-stranded DNA substrate. The number in the biotinylated nucleotide names refers to the carbon linker. ddATP-11-biotin, ddGTP-11-biotin, dATP-11-biotin, and dGTP-11-biotin were biotinylated from the 7-position of the purine base while dATP-14-biotin was biotinylated from 6-position of the purine base. Red circles indicate the migration of primer without elongation. B, Fractional content of the total sequenced reads for different miRXplore miRNA libraries with different input amounts, 3′ labeling nucleotide, and post-cDNA synthesis clean-up (i.e., either gel-based or streptavidin pull-down). C, Scatter plot of miRXplore miRNA library CV and 3′ precision of various biotinylated OTTR libraries. The Buffer 4A (either 4A, 4A1, or 4A4) used in each library is labeled by color of the data points. Buffer 4A recipes are specified at right. The shape of the data points indicates the nucleotide and reaction condition used during 3′ labeling. Libraries with the same 3′ labeling nucleotide and Buffer 4A are grouped together by a bounding line as a visual aid. D, Distributions of the log₂ CPM of miRXplore miRNA based on miRNA 3′ nucleotide for the libraries in C, with red for purine and blue for pyrimidine as indicated by the key. Library-wide CV for each library is given in the top-left corner of the plots. Color of CV text indicates which Buffer 4A variant was used. A representative gel-based cDNA purification of OTTR cDNA library was included at right.