Diverse bacterial transcriptomes and drug tolerance phenotypes are associated with expression of CD11c in the host cell. Flow sorting of infected macrophages on the basis of Mtb GFP expression before dual RNA-seq analysis of the bacterial transcriptomes establishes links between bacterial stress, induction of drug tolerance, and the expression levels of the macrophage surface marker CD11c. (a) PCA of the Mtb transcriptome for the four different samples: hspx′::GFP^high, hspx′::GFP^low (this manuscript), and Mtb in AM and IM (Pisu et al., 2020a). PCA reveals that both activation (hspx′::GFP^high, hspx′::GFP^low samples) and ontogeny of the host immune cell (Mtb in AM, Mtb in IM) play a role in shaping bacterial responses during infection. (b) Heatmap showing relative expression levels for the Mtb drug tolerance gene signature. (c) Heatmap showing relative expression levels for the DosR regulon in Mtb. (d) Boxplots showing expression levels (in log-normalized counts) for the genes related to the heme-iron signature in Mtb. hspx′::GFP^high bacteria overexpress a set of genes involved in heme-iron uptake and catabolism in agreement with the finding that most hspx′::GFP^high bacteria are contained in macrophages that are heme-loaded. (e) Violin plots showing expression levels (in log-normalized counts) of the genes part of the hspx operon. (f) Scatter plots showing staining levels (in log-normalized values) for the surface markers CD11c and SiglecF in infected cells. Host cells containing hspx′::GFP^high bacteria are visualized in red, while cells containing hspx′::GFP^low Mtb are displayed in light blue. Four different populations of host cells (SiglecF⁺/CD11c^low, SiglecF⁺/CD11c^high, SiglecF⁻/CD11c^int, and SiglecF⁻/CD11c^low), each containing different proportions of associated hspx′::GFP^high bacteria, are identifiable. The data clearly show that expression of the surface marker CD11c is inversely correlated with the bacterial phenotype in both AMs and IMs. (g) Violin plot showing staining levels (in log-normalized values) for the surface marker CD11c in hspx′::GFP^high- and hspx′::GFP^low- infected host cells. (h) Flow cytometry analysis of infected macrophages (MerTK⁺ CD64⁺ mCherry⁺) at 3 w.p.i. from hspx′::GFP/smyc′::mCherry infected mice. The same populations we identified by scRNA-seq (f) are also visible by flow cytometry. SSC-A, side scatter area. (i) Flow cytometry analysis of the MFI for the GFP signal across the different subpopulations of infected macrophages (AM CD11c^high, AM CD11c^low, IM CD11c^low, and IM CD11c^high). (j) Quantification of the amount of Mtb that survived exposure to 1 μg/ml of either INH or RIF in CD11c^high and CD11c^low macrophages. Values have been normalized for 100,000 bacteria in the untreated group. An 83% and 47% increase in the number of bacteria that survived drug treatment in CD11c^low macrophages was observed for INH and RIF, respectively. n = 5 for the infected population, n = 2 for the uninfected population, n = 3 for the bacterial transcriptome. The statistical significance for the genes part of the bacterial transcriptome has been calculated using the Wald test as implemented in the DESeq2 package (FDR < 0.05; see Materials and methods; Love et al., 2014). The statistical significance is provided for the remaining plots part of the flow cytometry and drug tolerance experiments (*, p-adj. < 0.05; **, p-adj. < 0.01; ***, p-adj. < 0.001; and ****, p-adj. < 0.0001; one-way ANOVA with Tukey test and unpaired t test; see Materials and methods).