Single-cell analysis reveals Mycobacterium tuberculosis ESX-1-mediated accumulation of permissive macrophages in infected mouse lungs

Abstract

Mycobacterium tuberculosis (MTB) ESX-1, a type VII secretion system, is a key virulence determinant contributing to MTB's survival within lung mononuclear phagocytes (MNPs), but its effect on MNP recruitment and differentiation remains unknown. Here, using multiple single-cell RNA sequencing techniques, we studied the role of ESX-1 in MNP heterogeneity and response in mice and murine bone marrow-derived macrophages (BMDM). We found that ESX-1 is required for MTB to recruit diverse MNP subsets with high MTB burden. Further, MTB induces a transcriptional signature of immune evasion in lung macrophages and BMDM in an ESX-1-dependent manner. Spatial transcriptomics revealed an up-regulation of permissive features within MTB lesions, where monocyte-derived macrophages concentrate near MTB-infected cells. Together, our findings suggest that MTB ESX-1 facilitates the recruitment and differentiation of MNPs, which MTB can infect and manipulate for survival. Our dataset across various models and methods could contribute to the broader understanding of recruited cell heterogeneity during MTB lung infection.

Fig. 1.. MTB induces recruitment of heterogeneous MNPs in the lungs of mice.

(A) Mice were infected with MTB for 28 days, lung dissociated to single-cell suspension, and live cells flow-sorted to obtain myeloid cells differentiated by infection status based on bacterial fluorescence expression (fig. S1). Cells were either sorted directly into 384-well plates (path 1; SS2: SmartSeq2) or washed and counted for 10X chemistry (path 2). After reverse transcription and decontamination, samples were removed from the BSL3 for library preparation and sequencing. (B) Ingest was used to combine SmartSeq2 and 10X sequencing datasets into a single UMAP plot. (C) Cellular subsets were classified using the Immgen database within SinglR for unbiased annotation. (D) Dot plot shows transcriptional canonical markers where fraction of cells expressing the gene is denoted by size of dot, and the darker red denotes higher mean gene expression of the gene. (E) Gene expression of key canonical markers after log transformation and normalization visualized using CELLxGENE, where darker red signifies higher expression. (F) Density plots of subsets obtained through index sorting for SmartSeq2 using the MFI of surface markers were converted using boolenization and overlayed onto the transcriptional data. Five mice per condition as biological replicates, with each condition replicated three times technically, for lung cell sorting.

Fig. 2.. SmartSeq2 reveals ESX-1 recruitment of diverse subsets of moDerived macrophages.

(A and B) C57BL/6 mice were infected for 28 days with H37Rv or H37RvΔEccD₁, and lung MNPs were analyzed by flow cytometry compared to uninfected control mice. (C) SmartSeq2 analysis of CD11b- and/or CD11c-positive cells. (D) Cellular subsets were further clustered using Leiden algorithm. (E) Separation by infection status of the mouse that the cells were isolated, with moDerived macrophages denoted by blue hashtag and transitional cells by an orange. Percent cell type make-up was determined for both (F) total mononuclear cells sorted per infection condition and (G) of all infected mononuclear cells per condition. **P < 0.01, ***P < 0.001, and ****P < 0.0001 using t test with Welch correction and Holm-Sidak multiple comparisons. For (A) and (B), flow cytometry analysis was performed with three to six mice per group for biological replicates with a minimum of two technical replicates across four separate experiments capturing a minimum of 100,000 events per condition per sort.

Fig. 3.. Recruited MNPs during lung MTB infection include a subset of cells with an anti-inflammatory signature.

(A) Differentially expressed marker genes identified in each cell cluster were used for Gene Ontology analysis with ShinyGOv0.77. In the dot plot, color indicates the negative log-transformed FDR, while dot size corresponds to the number of genes enriched in each pathway. (B) Gene sets were constructed to score the likelihood of a particular cell with transcriptional enrichment of the denoted pathway using the average expression of associated genes (Methods). Lighter color denotes higher expression of gene set and thus predicted attribute. moDer Macro, moDerived macrophage (Leiden clusters 1, 2, and 7); Trans Mono, transitioning monocyte (Leiden clusters 0, 3, and 10). (C) Violin plot of differentially expressed genes from the “anti-inflammatory” pathways (table S2) as delineated by Leiden cluster. (D) ZsGreen MFI of infected MNP subsets from H37Rv/ZsGreen-infected mouse lungs (28 days post infection). Ag, antigen; NO, nitric oxide; ROS, reactive oxygen species.

Fig. 4.. ESX-1 promotes macrophage activation and immune evasion response.

(A) Volcano plot of transitioning monocytes from mice infected with H37Rv and H37RvΔEccD₁ analyzed for differential gene expression differences using MAST. (B) Log₂ fold change differences of differentially expressed genes between H37Rv versus H37RvΔEccD₁ or each condition to cells present in the uninfected control mouse. (C) Volcano plot of differentially expressed genes in mature moDerived macrophages infected with H37Rv as compared to H37RvΔEccD₁. (D) Odds ratio for pathway enrichment using KEGG (mouse 2019), with the top 10 statistically significant pathways shown, removing redundancy. (E) QIAGEN IPA was also queried using differentially expressed genes determined from MAST analysis with an FDR < 0.1. Pathways shown are up-regulated, with a z score on the right x axis. The −log(FDR) is shown by red dots with values depicted on the left x axis. Fc, fold change.

Fig. 5.. ESX-1 aids in the suppression of the inflammatory response in BMDM.

(A) BMDM derived from C57BL/6 mice was infected with H37Rv or H37RvΔRD1 expressing dsRed for 24 hours, stained for viability, and flow-sorted on dsRed expression. Cells were multiplexed, and transcriptional libraries were obtained through 10X chemistry. (B) Dimension reduction of transcriptional libraries of each condition. (C) Heatmap of the top 20 differential gene expressions by log₂ fold change of each comparison: H37Rv or H37RvΔRD1 versus control and H37Rv versus H37RvΔRD1. (D) Pathway enrichment of differentially expressed genes of BMDM infected with H37Rv as compared to H37RvΔRD1. (E) Comparing differentially expressed genes up or downregulated in ESX-1 active or inactive strains from in vivo (moDerived macrophages) or in vitro (BMDM) (F) revealed 12 genes concurrent in differential expression. Each condition was repeated in triplicate, both biologically and technically. FC, fold change

Fig. 6.. MTB induces maturation of macrophages only near its presence.

(A) C57BL/6 mice were infected with H37Rv for 28 days, and after perfusion with PBS, lungs were inflated and freshly frozen in OCT. Slices (10 μm) were fixed and processed for Vizgen Merscope visualization of both MTB and gene panel. (B) Representative images from Merscope Visualizer. Red cell boundary indicates infected cells (i.e., MTB⁺), and green indicates uninfected cells (i.e., naïve). All circled cells express at least 20 transcripts from the panel. (C) Plot showing transcripts that are significantly up-regulated within 100 μm of MTB-infected cells. y axis is the difference between mean expression for a transcript in cells within 100 μm of MTB-infected cells as compared to that within all cells normalized to polyT (total mRNA per cell). All genes had a P value of <0.01 compared to pooled blank transcripts. (D) Violin plots showing the percent expression of up-regulated [shown in (C)], down-regulated (Lyve1, Sup1, Scarf2, Itga6, Jun, Icam2, and Car4), and unaffected genes within SmartSeq atlas shown in Fig. 4, with quartiles distinguished by lines within the violin plot. (E) Cumulative expression of representative markers for monocytes/macrophages. x axis represents the normalized gene counts from all listed MNP genes combined, and the y axis is the percentage of cells expressing less than or equal to the given expression level. A total of <1% of infected cells (black line) and <20% of all cells (gray line) have no detectable expression of the marker genes. (F) Representative cumulative distribution plots show a gene highly enriched near MTB-infected cells (Nos2), down-regulated near MTB-infected cells (Lyve1), and one that is unaffected by MTB presence (Apoe). Each color represents a bin of width 100 μm. Notably, not all lines start with y = 0 due to sparse gene expression levels. N = 3 uninfected, and N = 4 infected.