Research ArticleCELL BIOLOGY

Multiplexed single-cell RNA-seq via transient barcoding for simultaneous expression profiling of various drug perturbations

See allHide authors and affiliations

Science Advances  15 May 2019:
Vol. 5, no. 5, eaav2249
DOI: 10.1126/sciadv.aav2249
  • Fig. 1 Scheme and validation of transient barcoding method.

    (A) Scheme of multiplexed scRNA-seq by transient barcoding method using SBOs. (1) Samples with various conditions are prepared. (2) Each sample is transfected with SBO containing a unique sample barcode. (3) Barcoded cells are pooled together and processed for scRNA-seq (e.g., Drop-Seq). (4) Cells are lysed within droplets, and the released mRNAs and SBOs are captured, reverse-transcribed, and sequenced. (5) Cells are demultiplexed and assigned to their origins and processed for further analysis. (B) Heatmap of normalized SBO counts for 6-plex human/mouse species-mixing experiment. Rows represent cells, and columns represent SBOs. Cells are assessed whether they are positive for a particular SBO based on the SBO count matrix (see Materials and Methods). Cells were classified as singlets (positive for a unique SBO), multiplets (positive for more than one SBO), or negatives (not positive for any SBO) and ordered by their classifications. (C) Scatter plot showing raw counts between two SBOs. SBOs 1 and 6 were used to barcode different samples (Human 1, Mouse 2) (left). SBOs 3 and 4 were used to barcode the same sample (Human 3) (right). (D) Species-mixing plot of samples associated with SBOs 1 and 5. Cells were labeled according to their SBO classifications. Black dots indicate Human 1 sample barcoded with SBO 1, red dots indicate Mouse 1 sample barcoded with SBO 5, and gray dots indicate doublets that are positive for both SBOs. (E) Distribution of RNA transcript counts in cells between singlets (green), multiplets (blue), and negatives (red). Negatives, which imply beads exposed to ambient RNA, had the lowest number of transcripts. Multiplets had slightly more transcripts than singlets, indicating more RNA content within a droplet.

  • Fig. 2 Pseudotime analysis in 5-plex time-course experiment.

    (A) Monocle pseudotime trajectory of K562 cells treated with imatinib at different time points. Cells are labeled by pseudotime (top) and drug treatment time (bottom). The 0-, 6-, 12-, 24-, and 48-hour samples consist of 133, 109, 79, 49, 58, 52, and 90 cells, respectively. (B) Boxplot showing the distribution of pseudotime within each sample. (C) Prominent gene expression alterations in 5-plex time-course experiments of imatinib treatment. Note that the cells are labeled by drug treatment time and are not synchronously distributed over pseudotime. (D) Expression heatmap showing 50 genes with the lowest q values. (E) Expression heatmap showing DEGs between two transition states with q < 1 ×10−4. Prebranch refers to the cells before branch 1, Cell fate 1 refers to the cells of upper transition state, and Cell fate 2 refers to the cells in the lower transition state.

  • Fig. 3 Gene expression analysis in 48-plex drug treatment experiments.

    (A) Hierarchical clustered heatmap of averaged gene expression profiles for 48-plex drug treatment experiments in K562 cells. Each column represents averaged data in a drug, and each row represents a gene. DEGs were used in this heatmap. The scale bar of relative expression is on the right side. The ability of the drugs to inhibit kinase proteins is shown as binary colors (dark gray indicating positive) at the top. The bar plot at the top shows the cell count for each. (B) Volcano plot displaying DEGs of imatinib mesylate compared with DMSO controls. Genes that have a P value smaller than 0.05 and an absolute value of log (fold change) larger than 0.25 are considered significant. Up-regulated genes are colored in green, down-regulated genes are colored in red, and insignificant genes are colored in gray. Ten genes with the lowest P value are labeled. (C) Venn diagram showing the relationship between DEGs of three drug groups. Fourteen drugs are classified into three groups according to their protein targets (see Fig. 2C, top), and differential expression analysis is performed by comparing each group with DMSO controls. Relations of both positively (left) and negatively (right) regulated genes in each group are shown. (D) Plot showing a correlation between fold changes of expression in cells treated with mTOR inhibitors and BCR-ABL inhibitors compared with DMSO controls.

  • Fig. 4 Single-cell analysis in 48-plex drug treatment experiments.

    (A) The t-distributed stochastic neighbor embedding (t-SNE) plot of single cells in the 48-plex K562 samples. Plot shows six clusters (top), and additional t-SNE plot is labeled by cell cycle states (bottom). (B) Bar plots for 48-plex drug treatment experiments in K562 cells. The ability of the drugs to inhibit kinase proteins is shown as binary colors at the top (from Fig. 3A). The bar plot in the middle represents a relative fraction of cells in each t-SNE cluster [shown in (A)], and the bottom bar plot displays fractions of cell cycle states for every sample. Drugs are sorted by hierarchical clustering. (C) Expression heatmap showing the markers of the clusters. The numbers at the bottom represent cluster numbers. (D) Scaled expression of representative genes within the t-SNE plot. Intensity of the purple color determines expression levels, with higher intensity correlating with higher gene expression.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/5/eaav2249/DC1

    Fig. S1. Schematic figure of transient transfection using SBO.

    Fig. S2. Additional data for the species-mixing experiment.

    Fig. S3. SBO barcoding in heterogeneous cell samples and the effect of transient transfection.

    Fig. S4. Correlation heatmap of average gene expression across the drugs.

    Fig. S5. Volcano plots showing DEGs for each drug.

    Fig. S6. Cell cycle analysis and t-SNE at a single-cell resolution.

    Fig. S7. Expression heatmap for A375 drug screening experiment.

    Fig. S8. Comparison between previous methods and our method.

    Fig. S9. Cell numbers and sequencing depth for each experiment.

    Table S1. Oligos and drugs used in the drug screening experiment.

  • Supplementary Materials

    The PDF file includes:

    • Fig. S1. Schematic figure of transient transfection using SBO.
    • Fig. S2. Additional data for the species-mixing experiment.
    • Fig. S3. SBO barcoding in heterogeneous cell samples and the effect of transient transfection.
    • Fig. S4. Correlation heatmap of average gene expression across the drugs.
    • Fig. S5. Volcano plots showing DEGs for each drug.
    • Fig. S6. Cell cycle analysis and t-SNE at a single-cell resolution.
    • Fig. S7. Expression heatmap for A375 drug screening experiment.
    • Fig. S8. Comparison between previous methods and our method.
    • Fig. S9. Cell numbers and sequencing depth for each experiment.

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Table S1 (Microsoft Excel format). Oligos and drugs used in the drug screening experiment.

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article