Prevalent single cell transcriptomic profiling (scRNA-seq) mechods are mainly based on synthesis and enrichment of full-length double-stranded complementary DNA. These approaches are challenging to generate accurate quantification of transcripts when their abundance is low or their full-length amplifications are difficult.
Based on their previous finding that Tn5 transposase can directly cut-and-tag DNA/RNA hetero-duplexes, Peking University researchers have developed SHERRY2, a specifically optimized protocol for scRNA-seq without second strand cDNA synthesis. SHERRY2 is free of pre-amplification and eliminates the sequence-dependent bias. In comparison with other widely-used scRNA-seq methods, SHERRY2 exhibits significantly higher sensitivity and accuracy even for single nuclei. Besides, SHERRY2 is simple and robust, and can be easily scaled up to high-throughput experiments. When testing single lymphocytes and neuron nuclei, SHERRY2 not only obtained accurate countings of transcription factors and long non-coding RNAs, but also provided bias-free results that enriched genes in specific cellular components or functions, which outperformed other protocols. With a few thousand cells sequenced by SHERRY2, the researchers confirmed expression and dynamics of Myc in different cell types of germinal centers, which were previously only revealed by gene-specific amplification methods. SHERRY2 is able to provide high sensitivity, high accuracy, and high throughput for those applications that require high number of genes identified in each cell. It can reveal the subtle transcriptomic difference between cells and facilitate important biological discoveries.
The workflow and general performance of SHERRY2 on single cell RNA-seq
(A) The workflow of SHERRY2 for scRNA-seq. Poly(A) tailed RNA is firstly released from single cells and reverse transcribed. The resulting RNA/cDNA hetero-duplex is then tagmented by Tn5 transposome, followed by gap-repair and indexed PCR. Optionally, chromatin can be digested during lysis. The entire protocol is performed in one tube and takes 3 hours. (B) Gene number (FPKM>1) with SmartSeq2, SHERRY2 and SHERRY when subsampling reads to 0.1, 0.2, 0.4, 0.6, 0.8 and 1 million reads. (C) Pairwise correlation of gene expression within replicates for the three scRNA-seq protocols. The correlation R-value was calculated by a linear fitting model with normalized counts of overlapped genes. (D) Gene body coverage detected by the three scRNA-seq protocols. The gray region represents the standard deviation of the normalized depth among replicates. (E) Components of reads that were mapped to different regions of the genome using the three scRNA-seq protocols. The error bars show the standard deviation. (F) Gene expression correlation between single HEK293T cells and 200-ng RNA extracted from HEK293T cells. Single-cell data were acquired by the three scRNA-seq protocols. Bulk RNA results were acquired by the standard NEBNext protocol. The correlation R-value was calculated by a linear fitting model with normalized gene counts. The samples in (B-F) are single HEK293T cells. The p-values in (B, C, F) were calculated by the Mann-Whitney U test.