SUPPLEMENTARY INFORMATION

Transcription

1 In the format provided by the authors and unedited. SUPPLEMENTARY INFORMATION ARTICLE NUMBER: DOI: /NMICROBIOL Single cell RNA seq ties macrophage polarization to growth rate of intracellular Salmonella Antoine Emmanuel Saliba 1,2, Lei Li 2, Alexander J. Westermann 1, Silke Appenzeller 2, Daphne A. C. Stapels 3, Leon N. Schulte 1, Sophie Helaine 3, Jörg Vogel 1, 1RNA Biology group, Institute for Molecular Infection Biology, University of Würzburg, Josef Schneider Straße 2, D Würzburg, Germany 2Core Unit SysMed, University of Würzburg, Josef Schneider Straße 2, D Würzburg, Germany 3Section of Microbiology, Medical Research Council (MRC) Centre for Molecular Bacteriology and Infection, Imperial College London, Armstrong Road, London SW7 2AZ, UK Present address : Comprehensive Cancer Center Mainfranken, University of Würzburg, Am Hubland, D Würzburg Correspondance and material request : joerg.vogel@uni wuerzburg.de This file contains: Supplementary Figures 1 to 12 Supplementary references NATURE MICROBIOLOGY Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

2 Supplementary figures Supplementary Figure 1 Characterization of bacterial input for infection and of bystander cells. a, Representative histogram of GFP (left) and mcherry (right) measured by flow cytometry intensities of the S. Typhimurium SL1344 WT (up) and SL1344 pfccgi (down).

3 Total number of bacteria analyzed n= 100,000. b, Percentage of Salmonella expressing GFP and mcherry in WT, GFP+ and pfccgi strains. Error bars (S.D.) are calculated on three independent biological replicates. c, qrt PCR data of one bacterial and one mammalian house keeping gene applied to a bacterial culture (500 µl, OD600 of 2.0), naïve macrophages (10,000 cells), bystanders (10,000 cells) and infected macrophages (10,000 cells). Error bars (S.D.) are calculated on two independent biological replicates. Cq: quantification cycle.

4 Supplementary Figure 2 Intracellular bacterial growth over the course of the infection. Representative time course of flow cytometry scatter plots of non infected (mock) and challenged macrophages. Gate 1 captures naïve macrophages and bystanders, gate 2 infected macrophages with a single bacterium, and gate 3 macrophages infected with bacteria that have proliferated (number of cells displayed by panel: Mock: 3,950 cells; 0 h: 9,375 cells; 6 h: 5,627 cells; 12 h: 3,283 cells; 18 h: 5,460 cells).

5 Supplementary Figure 3 Technical assessment of single cell RNA seq. a, Mapped reads classification by their classes over all the cells that pass the initial filter (n=60). Error bars (S.D.) are calculated on all cells. b c, Assessment of the dynamic range and the sensitivity of the singlecell RNA seq protocol. Average normalized read counts (b) and average detection rate (that is, the probability to have a read count value above 0) (c) for the 92 ERCC RNA spike in as a function of the number of RNA molecules across all the single cells that pass the initial quality control (n=60 cells, Supplementary Table 1). d, Identification of biologically variable genes using spike ins to model the technical noise. Coefficient of variation (CV 2 ) is plotted against the read counts for all the naïve and challenged cells. In blue, average values for ERCC spike ins (blue square) are fitted to a parametrized model 1 (Brennecke et al, 2013, PMID: ) (solid blue curve). Variable genes (red circles) among all genes (black circles) are labeled,

6 respectively. Supporting that bone marrow macrophages are fully differentiated, we validated that cell cycle dependent genes (green dots) are not variable.

7 Supplementary Figure 4 Single cell clustering. a, t SNE analysis of all 60 cells. Each cell is represented as a dot and colors indicate cellular identities as inferred from the original FC (flow cytometry) gates. b, Gap statistic of k means clustering using the RaceID algorithm. The first local maximum (k=3, indicated with an arrow) provides a good estimate for the number of clusters that achieves optimal separation of the data into subpopulations. The data points and error bars refer to the average and standard deviation of gap statistics among 50 bootstrap samples.

8 Supplementary Figure 5 PAGODA analysis applied to naïve and challenged cells. The dendrogram shows the overall clustering of all 60 cells. Every cell is related to the flow cytometry sorting gate (and its correlated bacterial growth status) and is associated to its corresponding cell group as defined in Fig. 2b. The Cell PC score heatmap below reflects the top 4 aspects of heterogeneity (P<0,05) detected by PAGODA and every aspect is associated to GO terms noted as row labels. Finally, every heterogeneity aspect is associated with underlying gene sets.

9 Supplementary Figure 6 Principal component analysis (PCA) applied to macrophages with non growing and growing bacteria. PCA of all variable genes among macrophages with growing (14 cells) and non growing bacteria (16 cells) (as inferred from the original flow cytometry gates (Fig. 1b)) is performed. In this analysis, bystanders and naïve cells are not taken into account. The contribution of each cell (dot) to the first two dimensions is plotted with color referring to initial cellular identity.

10 Supplementary Figure 7 Single cell differential gene expression (SCDE) approach defines gene clusters specific for PCA isolated cell groups. Cell groups I, II and III isolated

11 from PCA analysis (Fig. 2b) were compared in pairwise manner using the SCDE method 2 (Kharchenko et al, 2014, PMID : ) using all genes that passed quality control (QC) (see Methods). All genes with P<0.01 ( threshold ) were selected for downstream analysis and represented in a heatmap (Fig. 2c). a, Clusters A and B are derived by comparing naïve cells versus challenged cells, labelled respectively cell groups I and II plus III on the PCA map (Fig. 2b). b, Clusters C and D are derived by comparing cell groups II and III identified on the PCA map (Fig. 2b).

12 Supplementary Figure 8 Bimodal/sporadic proinflammatory gene expression in group II. a, On the PCA map on dimensions (Dim) 1 and 2 of all naïve and challenged cells (Fig. 2b), single cell gene expression of Tlr2 and Nlrp3 are color coded showing preference expression in the group II. b, Violon plot showing the expression of Tnf, Nlrp3, and Cxcl10 in Group II (Fig. 2b; 28 cells).

13 Supplementary Figure 9 Gene expression pattern in M2 like cells. On the PCA map on dimensions (Dim) 1 and 2 of all naïve and challenged cells (Fig. 2b, 60 cells), single cell gene expression of Mrc1, Ccl8, Spp1, Id1, Timp1 and Gpr35 are color coded showing prefered expression in group III.

14 Supplementary Figure 10 Independent confirmation of population segregation. A replicate experiment of infection, cell sorting and single cell RNA seq was performed. PCA of all cells that passed the technical filter (18 naïve macrophages, 23 bystanders, 20 macrophages containing non growing bacteria, and 20 macrophages containing growing bacteria) is plotted on the first two dimensions.

15 Supplementary Figure 11 Kinetics of the levels of IL4RA (CD124) during infection. Macrophages were left uninfected or infected with Salmonella for 20 h. Cells were recovered, labeled for IL4RA (CD124) and analyzed by flow cytometry at 0, 2, 6, 10 or 20 h (n >30,000 cells). From 10 h onwards, cells were separated in the different macrophage populations containing growing or non growing bacteria. Error bars (S.D.) are calculated on two independent biological replicates.

16 Supplementary Figure 12 Macrophage polarization depends on bacterial growth. Macrophages were infected with Salmonella for 20 h prior to being recovered, labeled and analyzed by flow cytometry (n >30,000 cells). Cells were gated depending on the growth status of the bacteria and separated in cells containing multiple non growing bacteria or cells containing growing bacteria. Levels of detection of IL1B, ARG1, IL4RA (CD124) and CD86 were measured and compared in the different macrophage populations. Error bars (S.D.) are calculated on the indicated n independent biological replicates. A two tailed one sample t test was applied to obtain P.

17 Supplementary references 1 Brennecke, P. et al. Accounting for technical noise in single cell RNA seq experiments. Nat Methods 10, , (2013). 2 Kharchenko, P. V., Silberstein, L. & Scadden, D. T. Bayesian approach to single cell differential expression analysis. Nat Methods 11, , (2014).