scDual-Seq: mapping the gene regulatory program of Salmonella infection by host and pathogen single-cell RNA-sequencing

Abstract

The interaction between a pathogen and a host is a highly dynamic process in which both agents activate complex programs. Here, we introduce a single-cell RNA-sequencing method, scDual-Seq, that simultaneously captures both host and pathogen transcriptomes. We use it to study the process of infection of individual mouse macrophages with the intracellular pathogen Salmonella typhimurium. Among the infected macrophages, we find three subpopulations and we show evidence for a linear progression through these subpopulations, supporting a model in which these three states correspond to consecutive stages of infection.

Fig. 1

A single-cell RNA-sequencing approach to studying host–pathogen interaction. a Heterogeneity of outcomes of intracellular infection is due to both Salmonella and macrophage states. scDual-Seq simultaneously produces the transcriptome of both the host and the pathogen and allows the identification of cellular subpopulations during infection. b Schematic of the scDual-Seq method. Reverse transcription is primed using random hexamers, followed by RNase treatment and 3’ polyA tailing. The second strand is synthesized using the CEL-Seq2 barcoded primers (see “Methods”). The samples are pooled together before the complementary DNA (cDNA) undergoes linear amplification by in vitro transcription. The amplified RNA is then reverse transcribed using a random primer with an overhang of the sequence complementary to the Illumina 3’ adaptor. cDNA with both Illumina adaptors are selected by polymerase chain reaction and the DNA library is sequenced using paired-end Illumina sequencing. c Mean number of unique transcripts identified across five technical replicates, for mouse (black) and Salmonella (red). Circles and error bars represent the mean and standard deviation. d Plot between the expression of the two technical replicates of 10 pg mouse RNA and 10 pg Salmonella RNA. e Boxplots indicating the correlation coefficients across replicates with the sum expression of all 20 samples for mouse and for five replicates in each dilution for Salmonella. Mouse indicated in black, Salmonella dilutions indicated in red

Fig. 2

Identifying host subpopulations in mouse macrophages exposed to Salmonella. a Bone marrow-derived macrophages, exposed or unexposed to Salmonella, were sorted into a 96-well plate and processed using scDual-Seq at the four indicated time-points. b tSNE plot of single cells (perplexity = 10) computed based on correlation matrix between single cells using 457 mouse genes with high expression variation (mean/median > 1) and maximum expression higher than 10 tpm. The color indicates exposed (green), unexposed (gray), and induced (black circle). DBscan was used to cluster the cells into two groups. c Boxplot of expression levels (log₁₀ tpm) of the indicated mouse genes across the non-infected, partial-induced, and induced single cells (P < 0.02, Wilcoxon rank sum test). d Boxplots of the sum expression of the previously reported infection gene module [3] across the exposed and unexposed individual cells in our scDual-Seq data (P < 0.0001, Wilcoxon rank sum test). e tSNE plot of single cells (perplexity = 10) computed based on the normalized expression of the previously reported infection gene module [3]. Cells were colored according to their annotation in (a)

Fig. 3

A time-course of Salmonella infection. a tSNE plot of induced single cells (perplexity = 10) computed based upon 32 Salmonella regulons. Color indicates Class I (purple) and Class II (orange). DBscan was used to cluster the cells into two groups. b The heatmap indicates Pearson’s correlation coefficients between the average Salmonella regulons expression of the partially induced, Class I, and Class II subpopulations. c tSNE plot (positioned as in Fig. 2e) colored by the pseudo-time order of the cells. The bar below indicates unexposed (gray), partially induced (green), Class I (purple), and Class II (orange) cells, ordered by pseudo-time. d Plot of normalized expression level of SPI-1 and SPI-2 regulons in single cells ordered according to pseudo-time and smoothed (see “Methods”). The bar below indicates partially induced (green), Class I (purple), and Class II (orange) cells, ordered by pseudo-time. e Bar chart indicating for each time-point (2.5, 4, and 8 h after infection), the fraction of the three identified subpopulations (partially induced, Class I, and Class II)

Fig. 4

A model of consecutive infection stages. A model for the gene regulatory program of Salmonella infection. Host and pathogen transcriptomic processes are indicated in blue and red, respectively