Tn5 transposase and tagmentation procedures for massively scaled sequencing projects

Abstract

Massively parallel DNA sequencing of thousands of samples in a single machine-run is now possible, but the preparation of the individual sequencing libraries is expensive and time-consuming. Tagmentation-based library construction, using the Tn5 transposase, is efficient for generating sequencing libraries but currently relies on undisclosed reagents, which severely limits development of novel applications and the execution of large-scale projects. Here, we present simple and robust procedures for Tn5 transposase production and optimized reaction conditions for tagmentation-based sequencing library construction. We further show how molecular crowding agents both modulate library lengths and enable efficient tagmentation from subpicogram amounts of cDNA. The comparison of single-cell RNA-sequencing libraries generated using produced and commercial Tn5 demonstrated equal performances in terms of gene detection and library characteristics. Finally, because naked Tn5 can be annealed to any oligonucleotide of choice, for example, molecular barcodes in single-cell assays or methylated oligonucleotides for bisulfite sequencing, custom Tn5 production and tagmentation enable innovation in sequencing-based applications.

Figure 1.

Production and purification of tagmentation-ready Tn5. (A) Illustration of the Tn5–intein fusion construct. (B) 10% SDS-PAGE run with crude lysate (Lys), supernatant (Sup), flow-through (FT), wash (W), and DTT eluted fractions (DTT) from the chitin column. The expected sizes of the Tn5–intein fusion protein (Mxe-Tn5) and cleaved full-length Tn5 (Tn5) are indicated on the right; marker molecular weights in kDa, to the left. (C) Agarose gel (1% Type LE, in TAE buffer) demonstrating tagmentation of 50 or 100 ng high MW calf thymus DNA with 1 μL of Tn5 prepared by annealing in solution (Free), on chitin columns (Col), or with no Tn5 as a negative control (C). On-column annealing removed excess oligonucleotides present after solution annealing.

Figure 2.

Efficient and robust tagmentation reactions. (A) Bioanalyzer electropherograms of tagmented DNA libraries using Nextera (with column purification), in-house Tn5, and buffers with and without column purification. (B) Bioanalyzer electropherograms of tagmented DNA libraries generated with in-house Tn5 that were stripped with different amounts of SDS. (C) Fraction of sequenced reads mapping to DNA ladder sequences of varying lengths. The fraction of reads was adjusted for molarity of DNA ladder sequence, as determined with Bioanalyzer (Agilent).

Figure 3.

Comparable single-cell gene expression profiles with in-house Tn5. (A) Percentage of genes reproducibly detected in replicate cells, binned according to expression level. We performed all pairwise comparisons within replicates for the libraries generated with Nextera protocol or five modified protocols exchanging different procedure components (Supplemental Table 1) and report the mean and 90% confidence interval. (B) Standard deviation in gene expression estimates within replicates for Nextera or five variant protocols (as in A) in bins of genes sorted according to expression levels. Error bars, SEM (n ≥ 4). (C,D) Comparison, as in A and B, for sequence libraries generated with identical reaction buffers (Supplemental Table 1) but with either produced or commercial Nextera Tn5 transposase.

Figure 4.

Tagmentation reactions with molecular crowding agents. (A) Bioanalyzer electropherograms of tagmented DNA from reactions with different concentrations of PEG 8000 or no PEG or using Nextera XT reaction conditions. (B) Bioanalyzer electropherograms of tagmented DNA using PEG polymers of different MWs. (C) Representative Bioanalyzer electropherograms of tagmented DNA used to optimize the concentration of PEG 8000 in tagmentation reactions and to demonstrate how tagmented fragment size can be modulated by PEG amounts. (A–C) All reactions were performed on 100 pg HEK293T cell cDNA together with 12 (A) or 13 (B,C) cycles of enrichment PCR.

Figure 5.

Lower limit of starting cDNA for tagmentation libraries. (A,B) Bioanalyzer electropherograms of tagmented DNA from reactions using 500 pg to 0.1 pg cDNA using either Nextera XT (A) or in-house Tn5 and reaction conditions including TAPS buffer and 8% PEG 8000 (B). (C) Technical variability in transcriptome libraries generated with Nextera XT or in-house buffers (std, standard Tn5 amounts; low, 1/100th Tn5 amount). Mean and standard deviation in gene expression estimates within replicates per protocol and in bins of genes sorted according to expression levels. Error bars, SEM (n ≥ 3). (D) Sensitivity in transcriptome libraries generated with Nextera XT or in-house buffers (std, standard Tn5 amounts; low, 1/100th Tn5 amount). We report the mean percentage of genes reproducibly detected in replicate experiments (all pairwise comparisons per protocol) and binned according to expression level, with 90% confidence intervals.