ESX-5 Deletions in Mycobacterium tuberculosis Alter Macrophage Cytokine Signaling and Bacterial Heavy Metal Response

Abstract

The ESX-5 secretion system is critical for Mycobacterium tuberculosis (Mtb) viability and putatively linked to pathogenesis, but a functional understanding of how it interacts with the host is unknown. ESX-5 is encoded at a single genomic locus with small, paralogous, secreted targets (ESX-5a, 5b, 5c) spaced throughout the genome. To examine host-pathogen interactions of these putative virulence clusters, we made mutant strains lacking these loci and infected primary human macrophages. Surprisingly, all deletion mutants independently reduced cytokine secretion during infection, specific to certain analytes. This defect depended on viable bacteria and was mediated by a post-transcriptional mechanism. In bacterial transcriptomic analyses, each mutant downregulated heavy metal response genes compared to wild type bacteria. Treatment of Mtb with Cu or Cd led to increased ESX-5a and ESX-5c expression, concurrent with increased TNF and IL-6 secretion in macrophages compared to untreated bacilli, indicating a link between ESX-5 expression and cellular cytokine levels.

Figure 1 –. ESX-5 paralog deletion mutants induce differential inflammatory profiles in infected monocyte-derived macrophages.

1A. ESX-5 paralog deletions do not impact bacterial fitness in macrophages. 72hr replication levels via CFU (y-axis Log10 scale) recovered from N=6 infected donors. 1B. ESX-5 deletion growth is not attenuated in broth culture. Bar chart displays broth replication levels (y-axis OD600) for all knockout and wildtype strains (x-axis) for N=10 experiments. 1C. ESX-5 deletions alter phagolysosomal acidification. Bar chart displays acidic organelle staining in macrophages 6 hours post infection (6hpi) for N=3 biological donors (technical triplicate). Strain used on the x-axis (N.I. = uninfected control, No LT = no lysotracker control) and mean fluorescence intensity (MFI) on the y-axis. 1D. ESX-5 deletions induce altered cytokine profile in MDMs. Supernatant levels of TNF, IL-6, IL-1β (y-axis pg/ml) after Mtb infection of MDMs (x-axis strain or uninfected (NI)) (N=3 donors run in technical triplicate). 1E. Bacterial inactivation restores cytokine production. Supernatant levels of TNF, IL-6, IL-1β (y-axis pg/ml) from MDMs (N=3 donors in technical triplicate) infected with 4% PFA inactivated strains or uninfected (N.I.) (x-axis). 1A,1C-E. Statistics calculated using two-way ANOVA with a post-hoc T-test and Dunnett correction for multiple comparisons (95% confidence interval (CI), p≤0.05). 1B. Statistics calculated using one-way ANOVA with a post-hoc T-test and Dunnett correction for multiple comparisons (95% CI, p≤0.05). Error bars show standard deviation of the mean.

Figure 2 –. ESX-5 paralog deletions alter macrophage cytokine levels at post-transcriptional stage.

2A. Mtb macrophage uptake is unchanged across mutant strains. Bar plot display 4-hour phagocytosis levels of respective strains (x-axis) measured via CFU (y-axis, log10 scale) for N=3 donors run in technical triplicate. 2B. Mtb induced cytokine mRNA induction is comparable across strains. Bar charts display relative expression of TNF, IL-6, and IL-1β transcripts (y-axis, fold change expression) across N=3 donors (technical singlet) for respective strains (x-axis). 2C. Mtb-induced cytokine protein is not retained with the cellular fraction of infected cells. Dot plots represent TNF, IL-6, and IL-1β levels (y-axis pg/mL) for either secreted (Secreted) or cellular fractions (Cellular Lysates) of cells (N=3 donors, technical triplicate) infected with respective strains (x-axis, N.I = uninfected). 2D. Proteasome inhibition does not rescue cytokine protein levels. Dot plots display paired graphs with 2C with additional treatment of 10μM MG132 (proteasome inhibition). 2A-2D. Statistics generated using two-way ANOVA with post-hoc T-test and Dunnett correction for multiple comparisons (95% CI, p≤0.05). Error bars show standard deviation of the mean.

Figure 3 –. ESX-5 regulates Mtb-induced signaling in macrophages with cytokine specificity

3A. Mtb-induced macrophage IFN-β is downregulated with mutant infection compared to wild type. Bar charts display IFN-β response levels (y-axis, relative luminescence (RLU)) across N=3 donors (technical triplicate) and a bacterial dose curve (MOI 1, 5, or 10) comparing wildtype vs. respective knockout strains with an uninfected control (N.I.) (x-axis). 3B. IL-8 levels are unaffected by ESX-5 deletion mutants. Bar chart shows the level of IL-8 expression (y-axis, pg/mL) across N=3 donors (technical triplicate) infected with respective strains (x-axis, N.I. = uninfected). 3C. Chemokines display heterogenous expression profile. Bar charts display protein levels (y-axis, pg/mL) for four distinct chemokines (CCL3, CCL4, CCL5, and GM-CSF) across N=5 donors (technical singlet) for respective infection conditions (x-axis, N.I. = uninfected). CCL4 shows log transformed concentrations on y axis due to abnormal data across infection conditions (negative Shapiro-Wilk normality test). 3A-3C. Statistics for all analyses generated using two-way ANOVA with post-hoc T-test and Dunnett correction for multiple comparisons (95% CI, p≤0.05). Error bars show standard deviation of the mean.

Figure 4 –. Macrophage miRNA and UPR responses are not ESX-5 dependent.

4A and B. Mtb-induced macrophage microRNA expression is not ESX-5-dependent. Mtb inducible microRNA species (x-axis, number 1–130) in human macrophages (N=3, technical singlet) using nCounter (y-axis, absolute quantification) are shown. The included miRNA data has been baseline subtracted to remove low or inconsistently detected targets (removal of 697 probes, Dot shape correlates with respective strain (Circle = H37Rv, Square = Δ5a, Triangle = Δ5b, and Diamond = Δ5c). Black square shows highest abundance target for all samples (hsa-miR-4454+ hsa-miR-7975). 4B depicts microRNA hsa-miR-4454+ hsa-miR-7975 which had a marginal trend towards a difference between Δ5c and wild type. 4C. Mtb-induced macrophage UPR and ISR responses are not ESX-5-dependent. UPR expression targets HSPA5, TRIB3, and spliced XBP1 are shown with fold change differences (y-axis, (Mtb/Media)) across respective strain infections (x-axis, N.I. = uninfected). Dots represent N=5 biological donors run in technical singlet for HSPA5 and TRIB3 while technical duplicate for XBP1. 4B-C. Statistics generated using two-way ANOVA with post-hoc T-test and Dunnett correction for multiple comparisons (95% CI, p≤0.05). Error bars show standard deviation of the mean.

Figure 5 –. Bacterial transcriptional analysis of ESX-5 paralog knockouts reveals metal response axis.

5A. Volcano gene expression plots of mutants relative to wildtype. Plots show respective expression profiles across mutant transcriptomes relative to wildtype. For all plots, significance level is on y-axis (Log10 FDR Corrected (“Adjusted”) p-value) and change in expression on x-axis (Log2 Fold Change). Red dots denote significant genes (FDR ≤0.01, Log2FC ≥1/≤−1), grey dots are non-significant. 5B. String network of overlapped down-regulated genes shared across all knockout strains. Each node represents a gene while the connecting lines represent combined string score of respective nodes (minimum string score >0.700, medium high connectivity, string-db.org). 5C. Hypergeometric mean pathway enrichment analysis of overlapping genes. Enrichment plot shows top 7 most significant gene sets (y-axis) based on Gene Ontology (GO) biological process. X-axis shows the strength of the observation (Log10(observed genes/expected genes)). Size of node proportional to the number of genes in our data belonging to a respective gene set. Color of node correlates to the FDR of the enrichment (significance calculated using Fishers Exact test with Benjamini-Hochberg FDR correction).

Figure 6 –. Bacterial heavy metal treatment leads to differential Mtb ESX-5 paralog expression and increased cytokine output in macrophages.

6A. Mtb growth for heavy metal treated strains. Graph depicts 16 different growth curves (y-axis, OD600) (Paired strain (H37Rv = red, Δ5a = teal, Δ5b = blue, and Δ5c = pink) with respective metals (Iron 500μM = circle, Zinc 500μM = Triangle, Cadmium 100μM = Square, Copper 100μM = Diamond)) for 5 days (x-axis). Brackets on graph depict clustering of metal groups. 6B. Esx-5 paralog gene expression in H37Rv under metal stimulus. Bar charts show expression (y-axis, fold change expression) of a respective Esx gene cluster (Title) within H37Rv under different stimulation conditions (x-axis, 7H9= untreated control) across N=4 replicate experiments. Red dotted line indicates baseline expression of respective gene cluster in untreated bacteria. 6C. Metal Mtb pre-treatment leads to increased cytokine expression in macrophages. Bar charts show the relative abundance of TNF, IL-6, and IL-1β (y-axis, pg/mL) from N=5 donors (technical triplicate) infected with metal pre-treated H37Rv (x-axis, WT= no metal control, N.I. = uninfected). 6B-6C. Statistics generated using two-way ANOVA with post-hoc T-test and Dunnett correction for multiple comparisons (95% CI, p≤0.05). Error bars show standard deviation of the mean.