Comparative analysis of multiplex single-cell mRNA sequencing of resting and activated PBMCs using droplet-based and split-pool methods

Abstract

Single-cell mRNA sequencing is an essential technology for transcriptional profiling of cells and tissues. To compare transcriptomes among samples, it is cost-effective to multiplex their processing. Multiplexing is done by barcoding cDNA copies of transcripts from each sample, then combining them into a single library. We performed multiplex sequencing of human peripheral blood mononuclear cells (PBMCs) under three different experimental conditions: resting, treated with T cell activator, and treated with TNF-a. We generated libraries using two split-pool barcode ligation methods: Parse and a new in-house split-pool method, SWIFT-seq, developed by the Guttman group. PBMCs were fixed and permeabilized using a Parse kit for the Parse and SWIFT1 libraries. In a second experiment, we fixed and permeabilized cells using
a new in-house protocol and processed samples using SWIFT-seq to generate the SWIFT2 library. We also processed live cells using the Multiseq method for 10X Genomics-based sequencing, in which cells are indexed with lipid-linked barcodes, to generate the 10X library. These libraries encompass all major PBMC cell types, but myeloid cells were overrepresented in Parse and SWIFT1, probably due to the use of the Parse kit for fixation and permeabilization. Analysis of transcriptomes defined by these libraries shows that all sequencing and analysis methods generate remarkably similar biological conclusions. The Spearman rank correlation coefficients for comparisons of marker gene expression among all methods are >0.7 for all major PBMC cell types in resting and activated populations. The same gene programs are
implicated by the four methods as being involved in transitions from resting to activated states. Our results show that 10X, Parse, and SWIFT-seq methods can be used interchangeably for multiplex sequencing of PBMCs. We estimate that the per cell cost of analysis with SWIFT-seq, which uses reagents purchased individually from suppliers, is about one-third of that for the other methods.

Files

Additional details